Refactor to check HTTP content-type of request body.

src/allmydata/storage/http_server.py

@@ -2,7 +2,7 @@
 HTTP server for storage.
 """
 
-from typing import Dict, List, Set, Tuple
+from typing import Dict, List, Set, Tuple, Any
 
 from functools import wraps
 from base64 import b64decode
@@ -23,7 +23,7 @@ from werkzeug.datastructures import ContentRange
 from cbor2 import dumps, loads
 
 from .server import StorageServer
-from .http_common import swissnum_auth_header, Secrets
+from .http_common import swissnum_auth_header, Secrets, get_content_type
 from .common import si_a2b
 from .immutable import BucketWriter, ConflictingWriteError
 from ..util.hashutil import timing_safe_compare
@@ -248,7 +248,9 @@ class HTTPServer(object):
         return self._app.resource()
 
     def _send_encoded(self, request, data):
-        """Return encoded data, by default using CBOR."""
+        """
+        Return encoded data as the HTTP body response, by default using CBOR.
+        """
         cbor_mime = "application/cbor"
         accept_headers = request.requestHeaders.getRawHeaders("accept") or [cbor_mime]
         accept = parse_accept_header(accept_headers[0])
@@ -262,6 +264,18 @@ class HTTPServer(object):
             # https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3861
             raise _HTTPError(http.NOT_ACCEPTABLE)
 
+    def _read_encoded(self, request) -> Any:
+        """
+        Read encoded request body data, decoding it with CBOR by default.
+        """
+        content_type = get_content_type(request.requestHeaders)
+        if content_type == "application/cbor":
+            # TODO limit memory usage, client could send arbitrarily large data...
+            # https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3872
+            return loads(request.content.read())
+        else:
+            raise _HTTPError(http.UNSUPPORTED_MEDIA_TYPE)
+
     ##### Generic APIs #####
 
     @_authorized_route(_app, set(), "/v1/version", methods=["GET"])
@@ -280,7 +294,7 @@ class HTTPServer(object):
     def allocate_buckets(self, request, authorization, storage_index):
         """Allocate buckets."""
         upload_secret = authorization[Secrets.UPLOAD]
-        info = loads(request.content.read())
+        info = self._read_encoded(request)
 
         # We do NOT validate the upload secret for existing bucket uploads.
         # Another upload may be happening in parallel, with a different upload
@@ -480,6 +494,6 @@ class HTTPServer(object):
         except KeyError:
             raise _HTTPError(http.NOT_FOUND)
 
-        info = loads(request.content.read())
+        info = self._read_encoded(request)
         bucket.advise_corrupt_share(info["reason"].encode("utf-8"))
         return b""