Always use raw HTTP body (#316)

`RpcHttp.body` may be reformatted by the Functions Host, and as such,
may not be identical to the original body sent by the client.  This has
numerous problems (e.g invalid Content-Length etc.)  Sidestep the issue
by always using `RpcHttp.rawBody`.

See also: Azure/azure-functions-host#3875
Fixes: #260

Author: elprans

azure/functions_worker/bindings/http.py

@@ -47,10 +47,16 @@ def get_body(self) -> bytes:
         return self.__body_bytes
 
     def get_json(self) -> typing.Any:
-        if self.__body_type is meta.TypedDataKind.json:
+        if (self.__body_type is meta.TypedDataKind.json
+                or self.__body_type is meta.TypedDataKind.string):
             assert self.__body_str is not None
             return json.loads(self.__body_str)
-        raise ValueError('HTTP request does not have JSON data attached')
+        else:
+            try:
+                return json.loads(self.__body_bytes.decode())
+            except ValueError as e:
+                raise ValueError(
+                    'HTTP request does not contain valid JSON data') from e
 
 
 class HttpResponseConverter(meta.OutConverter, binding='http'):
@@ -105,7 +111,7 @@ def from_proto(cls, data: protos.TypedData, *,
         if data.WhichOneof('data') != 'http':
             raise NotImplementedError
 
-        body_rpc_val = data.http.body
+        body_rpc_val = data.http.rawBody
         body_rpc_type = body_rpc_val.WhichOneof('data')
 
         if body_rpc_type == 'json':

tests/http_functions/return_request/main.py

@@ -1,15 +1,18 @@
 import json
+import hashlib
 
 import azure.functions
 
 
 def main(req: azure.functions.HttpRequest):
     params = dict(req.params)
     params.pop('code', None)
+    body = req.get_body()
     return json.dumps({
         'method': req.method,
         'url': req.url,
         'headers': dict(req.headers),
         'params': params,
-        'get_body': req.get_body().decode(),
+        'get_body': body.decode(),
+        'body_hash': hashlib.sha256(body).hexdigest(),
     })

tests/test_http_functions.py

@@ -1,3 +1,5 @@
+import hashlib
+
 from azure.functions_worker import testutils
 
 
@@ -141,6 +143,18 @@ def test_post_return_request(self):
 
         self.assertEqual(req['get_body'], 'key=value')
 
+    def test_post_json_request_is_untouched(self):
+        body = b'{"foo":  "bar", "two":  4}'
+        body_hash = hashlib.sha256(body).hexdigest()
+        r = self.webhost.request(
+            'POST', 'return_request',
+            headers={'Content-Type': 'application/json'},
+            data=body)
+
+        self.assertEqual(r.status_code, 200)
+        req = r.json()
+        self.assertEqual(req['body_hash'], body_hash)
+
     def test_accept_json(self):
         r = self.webhost.request(
             'POST', 'accept_json',

tests/test_types.py

@@ -28,7 +28,7 @@ def test_http_request_bytes(self):
 
         self.assertEqual(r.get_body(), b'abc')
 
-        with self.assertRaisesRegex(ValueError, 'does not have JSON'):
+        with self.assertRaisesRegex(ValueError, 'does not contain valid JSON'):
             r.get_json()
 
         h = r.headers