Keep track of files put into zip package and don't skip their metadata even when they are zips

Author: Ben Greiner

dvuploader/file.py

@@ -29,6 +29,7 @@ class File(BaseModel):
 
     Private Attributes:
         _size (int): Size of the file in bytes.
+        _is_inside_zip (bool): Indicates if the file is packaged inside a zip archive.
 
     Methods:
         extract_file_name(): Extracts filename from filepath and initializes file handler.
@@ -57,6 +58,7 @@ class File(BaseModel):
     tab_ingest: bool = Field(default=True, alias="tabIngest")
 
     _size: int = PrivateAttr(default=0)
+    _is_inside_zip: bool = PrivateAttr(default=False)
 
     def extract_file_name(self):
         """

dvuploader/nativeupload.py

@@ -345,8 +345,8 @@ async def _update_metadata(
         try:
             if _tab_extension(dv_path) in file_mapping:
                 file_id = file_mapping[_tab_extension(dv_path)]
-            elif file.file_name and _is_zip(file.file_name):
-                # When the file is a zip it will be unpacked and thus
+            elif file.file_name and _is_zip(file.file_name) and not file._is_inside_zip:
+                # When the file is a zip package it will be unpacked and thus
                 # the expected file name of the zip will not be in the
                 # dataset, since it has been unpacked.
                 continue

dvuploader/packaging.py

@@ -98,6 +98,7 @@ def zip_files(
                 data=file.handler.read(),  # type: ignore
                 zinfo_or_arcname=_create_arcname(file),
             )
+            file._is_inside_zip = True
 
     return path