Merge pull request #710 from python-openapi/feature/request-response-binary-format-support

Request response binary format support

Author: p1c2u

Diff for: openapi_core/contrib/aiohttp/requests.py

@@ -19,7 +19,7 @@ class Empty:
 class AIOHTTPOpenAPIWebRequest:
     __slots__ = ("request", "parameters", "_get_body", "_body")
 
-    def __init__(self, request: web.Request, *, body: str | None):
+    def __init__(self, request: web.Request, *, body: bytes | None):
         if not isinstance(request, web.Request):
             raise TypeError(
                 f"'request' argument is not type of {web.Request.__qualname__!r}"
@@ -45,7 +45,7 @@ def method(self) -> str:
         return self.request.method.lower()
 
     @property
-    def body(self) -> str | None:
+    def body(self) -> bytes | None:
         return self._body
 
     @property

Diff for: openapi_core/contrib/aiohttp/responses.py

@@ -13,13 +13,13 @@ def __init__(self, response: web.Response):
         self.response = response
 
     @property
-    def data(self) -> str:
+    def data(self) -> bytes:
         if self.response.body is None:
-            return ""
+            return b""
         if isinstance(self.response.body, bytes):
-            return self.response.body.decode("utf-8")
+            return self.response.body
         assert isinstance(self.response.body, str)
-        return self.response.body
+        return self.response.body.encode("utf-8")
 
     @property
     def status_code(self) -> int:

Diff for: openapi_core/contrib/django/requests.py

@@ -76,9 +76,9 @@ def method(self) -> str:
         return self.request.method.lower()
 
     @property
-    def body(self) -> str:
+    def body(self) -> bytes:
         assert isinstance(self.request.body, bytes)
-        return self.request.body.decode("utf-8")
+        return self.request.body
 
     @property
     def content_type(self) -> str:

Diff for: openapi_core/contrib/django/responses.py

@@ -1,20 +1,28 @@
 """OpenAPI core contrib django responses module"""
+from itertools import tee
+
 from django.http.response import HttpResponse
+from django.http.response import StreamingHttpResponse
 from werkzeug.datastructures import Headers
 
 
 class DjangoOpenAPIResponse:
     def __init__(self, response: HttpResponse):
-        if not isinstance(response, HttpResponse):
+        if not isinstance(response, (HttpResponse, StreamingHttpResponse)):
             raise TypeError(
-                f"'response' argument is not type of {HttpResponse}"
+                f"'response' argument is not type of {HttpResponse} or {StreamingHttpResponse}"
             )
         self.response = response
 
     @property
-    def data(self) -> str:
+    def data(self) -> bytes:
+        if isinstance(self.response, StreamingHttpResponse):
+            resp_iter1, resp_iter2 = tee(self.response._iterator)
+            self.response.streaming_content = resp_iter1
+            content = b"".join(map(self.response.make_bytes, resp_iter2))
+            return content
         assert isinstance(self.response.content, bytes)
-        return self.response.content.decode("utf-8")
+        return self.response.content
 
     @property
     def status_code(self) -> int:

Diff for: openapi_core/contrib/falcon/requests.py

@@ -49,13 +49,14 @@ def method(self) -> str:
         return self.request.method.lower()
 
     @property
-    def body(self) -> Optional[str]:
+    def body(self) -> Optional[bytes]:
+        # Falcon doesn't store raw request stream.
+        # That's why we need to revert deserialized data
+
         # Support falcon-jsonify.
         if hasattr(self.request, "json"):
-            return dumps(self.request.json)
+            return dumps(self.request.json).encode("utf-8")
 
-        # Falcon doesn't store raw request stream.
-        # That's why we need to revert serialized data
         media = self.request.get_media(
             default_when_empty=self.default_when_empty,
         )
@@ -74,7 +75,7 @@ def body(self) -> Optional[str]:
             return None
         else:
             assert isinstance(body, bytes)
-            return body.decode("utf-8")
+            return body
 
     @property
     def content_type(self) -> str:

Diff for: openapi_core/contrib/falcon/responses.py

@@ -1,4 +1,6 @@
 """OpenAPI core contrib falcon responses module"""
+from itertools import tee
+
 from falcon.response import Response
 from werkzeug.datastructures import Headers
 
@@ -10,11 +12,16 @@ def __init__(self, response: Response):
         self.response = response
 
     @property
-    def data(self) -> str:
+    def data(self) -> bytes:
         if self.response.text is None:
-            return ""
+            if self.response.stream is None:
+                return b""
+            resp_iter1, resp_iter2 = tee(self.response.stream)
+            self.response.stream = resp_iter1
+            content = b"".join(resp_iter2)
+            return content
         assert isinstance(self.response.text, str)
-        return self.response.text
+        return self.response.text.encode("utf-8")
 
     @property
     def status_code(self) -> int:

Diff for: openapi_core/contrib/requests/requests.py

@@ -64,14 +64,14 @@ def method(self) -> str:
         return method and method.lower() or ""
 
     @property
-    def body(self) -> Optional[str]:
+    def body(self) -> Optional[bytes]:
         if self.request.body is None:
             return None
         if isinstance(self.request.body, bytes):
-            return self.request.body.decode("utf-8")
+            return self.request.body
         assert isinstance(self.request.body, str)
         # TODO: figure out if request._body_position is relevant
-        return self.request.body
+        return self.request.body.encode("utf-8")
 
     @property
     def content_type(self) -> str:

Diff for: openapi_core/contrib/requests/responses.py

@@ -10,9 +10,9 @@ def __init__(self, response: Response):
         self.response = response
 
     @property
-    def data(self) -> str:
+    def data(self) -> bytes:
         assert isinstance(self.response.content, bytes)
-        return self.response.content.decode("utf-8")
+        return self.response.content
 
     @property
     def status_code(self) -> int:

Diff for: openapi_core/contrib/starlette/requests.py

@@ -34,14 +34,14 @@ def method(self) -> str:
         return self.request.method.lower()
 
     @property
-    def body(self) -> Optional[str]:
+    def body(self) -> Optional[bytes]:
         body = self._get_body()
         if body is None:
             return None
         if isinstance(body, bytes):
-            return body.decode("utf-8")
+            return body
         assert isinstance(body, str)
-        return body
+        return body.encode("utf-8")
 
     @property
     def content_type(self) -> str:

Diff for: openapi_core/contrib/starlette/responses.py

@@ -1,20 +1,31 @@
 """OpenAPI core contrib starlette responses module"""
+from typing import Optional
+
 from starlette.datastructures import Headers
 from starlette.responses import Response
+from starlette.responses import StreamingResponse
 
 
 class StarletteOpenAPIResponse:
-    def __init__(self, response: Response):
+    def __init__(self, response: Response, data: Optional[bytes] = None):
         if not isinstance(response, Response):
             raise TypeError(f"'response' argument is not type of {Response}")
         self.response = response
 
+        if data is None and isinstance(response, StreamingResponse):
+            raise RuntimeError(
+                f"'data' argument is required for {StreamingResponse}"
+            )
+        self._data = data
+
     @property
-    def data(self) -> str:
+    def data(self) -> bytes:
+        if self._data is not None:
+            return self._data
         if isinstance(self.response.body, bytes):
-            return self.response.body.decode("utf-8")
+            return self.response.body
         assert isinstance(self.response.body, str)
-        return self.response.body
+        return self.response.body.encode("utf-8")
 
     @property
     def status_code(self) -> int:

Diff for: openapi_core/contrib/werkzeug/requests.py

@@ -39,8 +39,8 @@ def method(self) -> str:
         return self.request.method.lower()
 
     @property
-    def body(self) -> Optional[str]:
-        return self.request.get_data(as_text=True)
+    def body(self) -> Optional[bytes]:
+        return self.request.get_data(as_text=False)
 
     @property
     def content_type(self) -> str:

Diff for: openapi_core/contrib/werkzeug/responses.py

@@ -1,4 +1,6 @@
 """OpenAPI core contrib werkzeug responses module"""
+from itertools import tee
+
 from werkzeug.datastructures import Headers
 from werkzeug.wrappers import Response
 
@@ -10,8 +12,12 @@ def __init__(self, response: Response):
         self.response = response
 
     @property
-    def data(self) -> str:
-        return self.response.get_data(as_text=True)
+    def data(self) -> bytes:
+        if not self.response.is_sequence:
+            resp_iter1, resp_iter2 = tee(self.response.iter_encoded())
+            self.response.response = resp_iter1
+            return b"".join(resp_iter2)
+        return self.response.get_data(as_text=False)
 
     @property
     def status_code(self) -> int:

Diff for: openapi_core/deserializing/media_types/datatypes.py

@@ -2,5 +2,5 @@
 from typing import Callable
 from typing import Dict
 
-DeserializerCallable = Callable[[Any], Any]
+DeserializerCallable = Callable[[bytes], Any]
 MediaTypeDeserializersDict = Dict[str, DeserializerCallable]

Diff for: openapi_core/deserializing/media_types/deserializers.py

@@ -41,7 +41,9 @@ def __init__(
             extra_media_type_deserializers = {}
         self.extra_media_type_deserializers = extra_media_type_deserializers
 
-    def deserialize(self, mimetype: str, value: Any, **parameters: str) -> Any:
+    def deserialize(
+        self, mimetype: str, value: bytes, **parameters: str
+    ) -> Any:
         deserializer_callable = self.get_deserializer_callable(mimetype)
 
         try:
@@ -75,7 +77,7 @@ def __init__(
         self.encoding = encoding
         self.parameters = parameters
 
-    def deserialize(self, value: Any) -> Any:
+    def deserialize(self, value: bytes) -> Any:
         deserialized = self.media_types_deserializer.deserialize(
             self.mimetype, value, **self.parameters
         )
@@ -192,5 +194,4 @@ def decode_property_content_type(
                 value = location.getlist(prop_name)
                 return list(map(prop_deserializer.deserialize, value))
 
-        value = location[prop_name]
-        return prop_deserializer.deserialize(value)
+        return prop_deserializer.deserialize(location[prop_name])

Diff for: openapi_core/deserializing/media_types/exceptions.py

@@ -8,9 +8,9 @@ class MediaTypeDeserializeError(DeserializeError):
     """Media type deserialize operation error"""
 
     mimetype: str
-    value: str
+    value: bytes
 
     def __str__(self) -> str:
         return (
             "Failed to deserialize value with {mimetype} mimetype: {value}"
-        ).format(value=self.value, mimetype=self.mimetype)
+        ).format(value=self.value.decode("utf-8"), mimetype=self.mimetype)

Diff for: openapi_core/deserializing/media_types/util.py

@@ -10,16 +10,11 @@
 from werkzeug.datastructures import ImmutableMultiDict
 
 
-def binary_loads(value: Union[str, bytes], **parameters: str) -> bytes:
-    charset = "utf-8"
-    if "charset" in parameters:
-        charset = parameters["charset"]
-    if isinstance(value, str):
-        return value.encode(charset)
+def binary_loads(value: bytes, **parameters: str) -> bytes:
     return value
 
 
-def plain_loads(value: Union[str, bytes], **parameters: str) -> str:
+def plain_loads(value: bytes, **parameters: str) -> str:
     charset = "utf-8"
     if "charset" in parameters:
         charset = parameters["charset"]
@@ -32,30 +27,36 @@ def plain_loads(value: Union[str, bytes], **parameters: str) -> str:
     return value
 
 
-def json_loads(value: Union[str, bytes], **parameters: str) -> Any:
+def json_loads(value: bytes, **parameters: str) -> Any:
     return loads(value)
 
 
-def xml_loads(value: Union[str, bytes], **parameters: str) -> Element:
-    return fromstring(value)
+def xml_loads(value: bytes, **parameters: str) -> Element:
+    charset = "utf-8"
+    if "charset" in parameters:
+        charset = parameters["charset"]
+    return fromstring(value.decode(charset))
 
 
-def urlencoded_form_loads(value: Any, **parameters: str) -> Mapping[str, Any]:
-    return ImmutableMultiDict(parse_qsl(value))
+def urlencoded_form_loads(
+    value: bytes, **parameters: str
+) -> Mapping[str, Any]:
+    # only UTF-8 is conforming
+    return ImmutableMultiDict(parse_qsl(value.decode("utf-8")))
 
 
-def data_form_loads(
-    value: Union[str, bytes], **parameters: str
-) -> Mapping[str, Any]:
-    if isinstance(value, bytes):
-        value = value.decode("ASCII", errors="surrogateescape")
+def data_form_loads(value: bytes, **parameters: str) -> Mapping[str, Any]:
+    charset = "ASCII"
+    if "charset" in parameters:
+        charset = parameters["charset"]
+    decoded = value.decode(charset, errors="surrogateescape")
     boundary = ""
     if "boundary" in parameters:
         boundary = parameters["boundary"]
     parser = Parser()
     mimetype = "multipart/form-data"
     header = f'Content-Type: {mimetype}; boundary="{boundary}"'
-    text = "\n\n".join([header, value])
+    text = "\n\n".join([header, decoded])
     parts = parser.parsestr(text, headersonly=False)
     return ImmutableMultiDict(
         [

Diff for: openapi_core/protocols.py

@@ -17,7 +17,7 @@ def method(self) -> str:
         ...
 
     @property
-    def body(self) -> Optional[str]:
+    def body(self) -> Optional[bytes]:
         ...
 
     @property
@@ -120,7 +120,7 @@ class Response(Protocol):
     """
 
     @property
-    def data(self) -> str:
+    def data(self) -> Optional[bytes]:
         ...
 
     @property

Diff for: openapi_core/testing/requests.py

@@ -20,7 +20,7 @@ def __init__(
         view_args: Optional[Dict[str, Any]] = None,
         headers: Optional[Dict[str, Any]] = None,
         cookies: Optional[Dict[str, Any]] = None,
-        data: Optional[str] = None,
+        data: Optional[bytes] = None,
         content_type: str = "application/json",
     ):
         self.host_url = host_url