tusd: Remove non-ASCII characters from path-ids.

zerver/lib/upload/__init__.py

@@ -114,20 +114,27 @@ def get_public_upload_root_url() -> str:
     return upload_backend.get_public_upload_root_url()
 
 
-def sanitize_name(value: str) -> str:
-    """
-    Sanitizes a value to be safe to store in a Linux filesystem, in
+def sanitize_name(value: str, *, strict: bool = False) -> str:
+    """Sanitizes a value to be safe to store in a Linux filesystem, in
     S3, and in a URL.  So Unicode is allowed, but not special
     characters other than ".", "-", and "_".
 
+    In "strict" mode, it does not allow Unicode, allowing only ASCII
+    [A-Za-z0-9_] as word characters.  This is for the benefit of tusd,
+    which is not Unicode-aware.
+
     This implementation is based on django.utils.text.slugify; it is
     modified by:
     * adding '.' to the list of allowed characters.
     * preserving the case of the value.
     * not stripping trailing dashes and underscores.
+
     """
-    value = unicodedata.normalize("NFKC", value)
-    value = re.sub(r"[^\w\s.-]", "", value).strip()
+    if strict:
+        value = re.sub(r"[^A-Za-z0-9_ .-]", "", value).strip()
+    else:
+        value = unicodedata.normalize("NFKC", value)
+        value = re.sub(r"[^\w\s.-]", "", value).strip()
     value = re.sub(r"[-\s]+", "-", value)
 
     # Django's MultiPartParser never returns files named this, but we

zerver/tests/test_tusd.py

@@ -182,7 +182,7 @@ class TusdPreCreateTest(ZulipTestCase):
     def test_sanitize_filename(self) -> None:
         self.login("hamlet")
         request = self.request()
-        request.event.upload.meta_data["filename"] = "some thing! ... like this?"
+        request.event.upload.meta_data["filename"] = "some 例 thing! ... like this?"
         result = self.client_post(
             "/api/internal/tusd",
             request.model_dump(),

zerver/views/tusd.py

@@ -111,7 +111,7 @@ def handle_upload_pre_create_hook(
         return reject_upload(str(e), 413)
 
     # Determine the path_id to store it at
-    file_name = sanitize_name(data.meta_data.get("filename", ""))
+    file_name = sanitize_name(data.meta_data.get("filename", ""), strict=True)
     path_id = upload_backend.generate_message_upload_path(str(user_profile.realm_id), file_name)
     return tusd_json_response({"ChangeFileInfo": {"ID": path_id}})