re

cmdlet/download_file.py

@@ -319,6 +319,18 @@ class Download_File(Cmdlet):
                     except Exception:
                         provider_name = None
 
+                # Heuristic: LibGen often uses landing pages like edition.php/file.php.
+                # These should never be treated as direct file URLs.
+                if not provider_name:
+                    try:
+                        p = urlparse(str(url))
+                        h = (p.hostname or "").strip().lower()
+                        path = (p.path or "").strip().lower()
+                        if "libgen" in h and any(x in path for x in ("/edition.php", "/file.php", "/ads.php", "/get.php", "/series.php")):
+                            provider_name = "libgen"
+                    except Exception:
+                        pass
+
                 if provider_name and get_provider is not None and SearchResult is not None:
                     # OpenLibrary URLs should be handled by the OpenLibrary provider.
                     if provider_name == "openlibrary":
@@ -391,11 +403,19 @@ class Download_File(Cmdlet):
                         progress.clear_status()
 
                         if downloaded_path:
+                            tags_hint: Optional[List[str]] = None
+                            try:
+                                sr_tags = getattr(sr, "tag", None)
+                                if isinstance(sr_tags, set) and sr_tags:
+                                    tags_hint = sorted([str(t) for t in sr_tags if t])
+                            except Exception:
+                                tags_hint = None
+
                             self._emit_local_file(
                                 downloaded_path=Path(downloaded_path),
                                 source=str(url),
                                 title_hint=title_hint,
-                                tags_hint=None,
+                                tags_hint=tags_hint,
                                 media_kind_hint="book",
                                 full_metadata=sr.full_metadata,
                                 provider_hint="openlibrary",
@@ -472,14 +492,42 @@ class Download_File(Cmdlet):
                         except Exception:
                             downloaded_path = None
 
+                        # Refuse to fall back to direct-download for LibGen landing pages.
+                        # This prevents saving HTML (e.g. edition.php) as a bogus file.
+                        if (not downloaded_path) and str(provider_name).lower() == "libgen":
+                            raise DownloadError("LibGen URL did not resolve to a downloadable file")
+
                         if downloaded_path:
+                            tags_hint: Optional[List[str]] = None
+                            full_md: Optional[Dict[str, Any]] = None
+                            title_hint = Path(str(downloaded_path)).stem
+                            media_kind_hint = "file"
+
+                            if str(provider_name).lower() == "libgen":
+                                media_kind_hint = "book"
+                                try:
+                                    sr_tags = getattr(sr, "tag", None)
+                                    if isinstance(sr_tags, set) and sr_tags:
+                                        tags_hint = sorted([str(t) for t in sr_tags if t])
+                                except Exception:
+                                    tags_hint = None
+
+                                try:
+                                    if isinstance(getattr(sr, "full_metadata", None), dict):
+                                        full_md = sr.full_metadata
+                                        t = str(full_md.get("title") or "").strip()
+                                        if t:
+                                            title_hint = t
+                                except Exception:
+                                    full_md = None
+
                             self._emit_local_file(
                                 downloaded_path=Path(downloaded_path),
                                 source=str(url),
-                                title_hint=Path(str(downloaded_path)).stem,
-                                tags_hint=None,
-                                media_kind_hint="file",
-                                full_metadata=None,
+                                title_hint=title_hint,
+                                tags_hint=tags_hint,
+                                media_kind_hint=media_kind_hint,
+                                full_metadata=full_md,
                                 provider_hint=str(provider_name),
                                 progress=progress,
                                 config=config,
@@ -600,6 +648,7 @@ class Download_File(Cmdlet):
                 # If this looks like a provider item and providers are available, prefer provider.download()
                 downloaded_path: Optional[Path] = None
                 attempted_provider_download = False
+                provider_sr = None
                 if table and get_search_provider and SearchResult:
                     provider = get_search_provider(str(table), config)
                     if provider is not None:
@@ -612,6 +661,7 @@ class Download_File(Cmdlet):
                         )
                         debug(f"[download-file] Downloading provider item via {table}: {sr.title}")
                         downloaded_path = provider.download(sr, final_output_dir)
+                        provider_sr = sr
 
                 # OpenLibrary: if provider download failed, do NOT try to download the OpenLibrary page HTML.
                 if downloaded_path is None and attempted_provider_download and str(table or "").lower() == "openlibrary":
@@ -693,6 +743,30 @@ class Download_File(Cmdlet):
                     log(f"Cannot download item (no provider handler / unsupported target): {title or target}", file=sys.stderr)
                     continue
 
+                # Allow providers to add/enrich tags and metadata during download.
+                if str(table or "").lower() == "libgen" and provider_sr is not None:
+                    try:
+                        sr_tags = getattr(provider_sr, "tag", None)
+                        if tags_list is None and isinstance(sr_tags, set) and sr_tags:
+                            tags_list = sorted([str(t) for t in sr_tags if t])
+                    except Exception:
+                        pass
+
+                    try:
+                        sr_md = getattr(provider_sr, "full_metadata", None)
+                        if isinstance(sr_md, dict) and sr_md:
+                            full_metadata = sr_md
+                    except Exception:
+                        pass
+
+                    try:
+                        if isinstance(full_metadata, dict):
+                            t = str(full_metadata.get("title") or "").strip()
+                            if t:
+                                title = t
+                    except Exception:
+                        pass
+
                 self._emit_local_file(
                     downloaded_path=downloaded_path,
                     source=str(target) if target else None,