Open files just-in-time to prevent OSError

dvuploader/directupload.py

@@ -250,7 +250,7 @@ async def _upload_singlepart(
         "headers": headers,
         "url": ticket["url"],
         "content": upload_bytes(
-            file=file.handler,  # type: ignore
+            file=file.get_handler(),  # type: ignore
             progress=progress,
             pbar=pbar,
             hash_func=file.checksum._hash_fun,

dvuploader/file.py

@@ -78,7 +78,6 @@ def extract_file_name(self):
 
         if self.handler is None:
             self._validate_filepath(self.filepath)
-            self.handler = open(self.filepath, "rb")
             self._size = os.path.getsize(self.filepath)
         else:
             self._size = len(self.handler.read())
@@ -95,6 +94,15 @@ def extract_file_name(self):
 
         return self
 
+    def get_handler(self) -> IO:
+        """
+        Opens the file and initializes the file handler.
+        """
+        if self.handler is not None:
+            return self.handler
+
+        return open(self.filepath, "rb")
+
     @staticmethod
     def _validate_filepath(path):
         """
@@ -126,7 +134,6 @@ def apply_checksum(self):
 
         self.checksum.apply_checksum()
 
-
     def update_checksum_chunked(self, blocksize=2**20):
         """Updates the checksum with data read from a file-like object in chunks.
 
@@ -139,12 +146,13 @@ def update_checksum_chunked(self, blocksize=2**20):
         Note:
             This method resets the file position to the start after reading.
         """
-        assert self.handler is not None, "File handler is not initialized."
         assert self.checksum is not None, "Checksum is not initialized."
         assert self.checksum._hash_fun is not None, "Checksum hash function is not set."
 
+        handler = self.get_handler()
+
         while True:
-            buf = self.handler.read(blocksize)
+            buf = handler.read(blocksize)
 
             if not isinstance(buf, bytes):
                 buf = buf.encode()
@@ -153,8 +161,12 @@ def update_checksum_chunked(self, blocksize=2**20):
                 break
             self.checksum._hash_fun.update(buf)
 
-        self.handler.seek(0)
-
+        if self.handler is not None:  # type: ignore
+            # In case of passed handler, we need to seek the handler to the start after reading.
+            self.handler.seek(0)
+        else:
+            # Path-based handlers will be opened just-in-time, so we can close it.
+            handler.close()
 
     def __del__(self):
         if self.handler is not None:

dvuploader/nativeupload.py

@@ -92,8 +92,12 @@ async def native_upload(
     }
 
     files_new = [file for file in files if not file.to_replace]
-    files_new_metadata = [file for file in files if file.to_replace and file._unchanged_data]
-    files_replace = [file for file in files if file.to_replace and not file._unchanged_data]
+    files_new_metadata = [
+        file for file in files if file.to_replace and file._unchanged_data
+    ]
+    files_replace = [
+        file for file in files if file.to_replace and not file._unchanged_data
+    ]
 
     # These are not in a package but need a metadtata update, ensure even for zips
     for file in files_new_metadata:
@@ -114,7 +118,7 @@ async def native_upload(
                         file.file_name,  # type: ignore
                         total=file._size,
                     ),
-                    file
+                    file,
                 )
                 for file in files_replace
             ]
@@ -269,11 +273,12 @@ async def _single_native_upload(
         )
 
     json_data = _get_json_data(file)
+    handler = file.get_handler()
 
     files = {
         "file": (
             file.file_name,
-            file.handler,
+            handler,
             file.mimeType,
         ),
         "jsonData": (
@@ -374,8 +379,10 @@ async def _update_metadata(
             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)
-                and not file._is_inside_zip and not file._enforce_metadata_update
+                file.file_name
+                and _is_zip(file.file_name)
+                and not file._is_inside_zip
+                and not file._enforce_metadata_update
             ):
                 # 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

dvuploader/packaging.py

@@ -95,7 +95,7 @@ def zip_files(
     with zipfile.ZipFile(path, "w") as zip_file:
         for file in files:
             zip_file.writestr(
-                data=file.handler.read(),  # type: ignore
+                data=file.get_handler().read(),  # type: ignore
                 zinfo_or_arcname=_create_arcname(file),
             )
             file._is_inside_zip = True