Stage uploads on the cache filesystem

.changeset/seven-bears-stick.md

@@ -0,0 +1,5 @@
+---
+"gradio": patch
+---
+
+fix:Stage uploads on the cache filesystem

gradio/route_utils.py

@@ -686,7 +686,8 @@ class GradioMultiPartParser:
 
     Made the following modifications
         - Use GradioUploadFile instead of UploadFile
-        - Use NamedTemporaryFile instead of SpooledTemporaryFile
+        - Use NamedTemporaryFile instead of SpooledTemporaryFile, optionally
+          placing it in Gradio's upload directory
         - Compute hash of data as the request is streamed
 
     """
@@ -700,6 +701,7 @@ def __init__(
         *,
         max_files: Union[int, float] = 1000,
         max_fields: Union[int, float] = 1000,
+        upload_dir: str | Path | None = None,
         upload_id: str | None = None,
         upload_progress: FileUploadProgress | None = None,
         max_file_size: int | float,
@@ -709,6 +711,7 @@ def __init__(
         self.stream = stream
         self.max_files = max_files
         self.max_fields = max_fields
+        self.upload_dir = upload_dir
         self.items: list[tuple[str, Union[str, UploadFile]]] = []
         self.upload_id = upload_id
         self.upload_progress = upload_progress
@@ -804,7 +807,7 @@ def on_headers_finished(self) -> None:
                     f"Too many files. Maximum number of files is {self.max_files}."
                 )
             filename = _user_safe_decode(options[b"filename"], str(self._charset))
-            tempfile = NamedTemporaryFile(delete=False)
+            tempfile = NamedTemporaryFile(delete=False, dir=self.upload_dir)
             self._files_to_close_on_error.append(tempfile)
             self._current_part.file = GradioUploadFile(
                 file=tempfile,  # type: ignore[arg-type]
@@ -1542,6 +1545,7 @@ async def upload_fn(
     if content_type != b"multipart/form-data":
         raise HTTPException(status_code=400, detail="Invalid content type.")
 
+    Path(upload_dir).mkdir(exist_ok=True, parents=True)
     if upload_id and upload_progress:
         upload_progress.track(upload_id)
 
@@ -1550,6 +1554,7 @@ async def upload_fn(
         request.stream(),
         max_files=1000,
         max_fields=1000,
+        upload_dir=upload_dir,
         max_file_size=max_file_size,
         upload_id=upload_id,
         upload_progress=upload_progress,

test/test_routes.py

@@ -144,6 +144,38 @@ def test_custom_upload_path(self, gradio_temp_dir):
         with open(file, "rb") as saved_file:
             assert saved_file.read() == b"abcdefghijklmnopqrstuvwxyz"
 
+    def test_upload_is_staged_in_custom_upload_path(self, gradio_temp_dir, monkeypatch):
+        blocks = Blocks()
+        blocks.max_file_size = None
+        blocks.upload_file_set = set()
+        blocks.share = False
+        app = routes.App.create_app(blocks)
+        test_client = TestClient(app)
+        original_rename = os.rename
+        staged_paths = []
+
+        def record_rename(source, destination):
+            source = Path(source).resolve()
+            staged_paths.append(source)
+            if not source.is_relative_to(gradio_temp_dir.resolve()):
+                raise OSError("simulated cross-filesystem rename")
+            return original_rename(source, destination)
+
+        def reject_background_move(*_args):
+            raise AssertionError("upload must be published before the response")
+
+        monkeypatch.setattr(os, "rename", record_rename)
+        monkeypatch.setattr(
+            routes, "move_uploaded_files_to_cache", reject_background_move
+        )
+        with open("test/test_files/alphabet.txt", "rb") as file:
+            response = test_client.post(f"{API_PREFIX}/upload", files={"files": file})
+
+        assert response.status_code == 200
+        assert len(staged_paths) == 1
+        assert staged_paths[0].is_relative_to(gradio_temp_dir.resolve())
+        assert Path(response.json()[0]).read_bytes() == b"abcdefghijklmnopqrstuvwxyz"
+
     @pytest.mark.skipif(
         sys.platform == "win32",
         reason="On Windows CI python_multipart raises MultipartParseError while "