jkj

cmdlet/download_file.py

@@ -17,15 +17,15 @@ from SYS.download import DownloadError, _download_direct_file
 from SYS.logger import log, debug
 import pipeline as pipeline_context
 
-from ._shared import (
-    Cmdlet,
-    CmdletArg,
-    SharedArgs,
-    parse_cmdlet_args,
-    register_url_with_local_library,
-    coerce_to_pipe_object,
-    get_field,
-)
+from . import _shared as sh
+
+Cmdlet = sh.Cmdlet
+CmdletArg = sh.CmdletArg
+SharedArgs = sh.SharedArgs
+parse_cmdlet_args = sh.parse_cmdlet_args
+register_url_with_local_library = sh.register_url_with_local_library
+coerce_to_pipe_object = sh.coerce_to_pipe_object
+get_field = sh.get_field
 
 
 class Download_File(Cmdlet):
@@ -251,6 +251,13 @@ class Download_File(Cmdlet):
 
                     # Fallback: if we have a direct HTTP URL, download it directly
                     if downloaded_path is None and isinstance(target, str) and target.startswith("http"):
+                        # Guard: provider landing pages (e.g. LibGen ads.php) are HTML, not files.
+                        # Never download these as "files".
+                        if str(table or "").lower() == "libgen":
+                            low = target.lower()
+                            if ("/ads.php" in low) or ("/file.php" in low) or ("/index.php" in low):
+                                log("[download-file] Refusing to download LibGen landing page (expected provider to resolve file link)", file=sys.stderr)
+                                continue
                         debug(f"[download-file] Provider item looks like direct URL, downloading: {target}")
                         result_obj = _download_direct_file(target, final_output_dir, quiet=quiet_mode)
                         file_path = None