Add http and restjson1 client protocols (smithy-lang#448)

Author: haydenbaker

packages/smithy-aws-core/src/smithy_aws_core/aio/__init__.py

@@ -0,0 +1,2 @@
+#  Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#  SPDX-License-Identifier: Apache-2.0

packages/smithy-aws-core/src/smithy_aws_core/aio/protocols.py

@@ -0,0 +1,27 @@
+from typing import Final
+
+from smithy_aws_core.traits import RestJson1Trait
+from smithy_http.aio.protocols import HttpBindingClientProtocol
+from smithy_core.codecs import Codec
+from smithy_core.shapes import ShapeID
+from smithy_json import JSONCodec
+
+
+class RestJsonClientProtocol(HttpBindingClientProtocol):
+    """An implementation of the aws.protocols#restJson1 protocol."""
+
+    _id: ShapeID = RestJson1Trait.id
+    _codec: JSONCodec = JSONCodec()
+    _contentType: Final = "application/json"
+
+    @property
+    def id(self) -> ShapeID:
+        return self._id
+
+    @property
+    def payload_codec(self) -> Codec:
+        return self._codec
+
+    @property
+    def content_type(self) -> str:
+        return self._contentType

packages/smithy-aws-core/src/smithy_aws_core/traits.py

@@ -0,0 +1,44 @@
+#  Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#  SPDX-License-Identifier: Apache-2.0
+
+# This ruff check warns against using the assert statement, which can be stripped out
+# when running Python with certain (common) optimization settings. Assert is used here
+# for trait values. Since these are always generated, we can be fairly confident that
+# they're correct regardless, so it's okay if the checks are stripped out.
+# ruff: noqa: S101
+
+from dataclasses import dataclass, field
+from typing import Mapping, Sequence
+
+from smithy_core.shapes import ShapeID
+from smithy_core.traits import Trait, DocumentValue, DynamicTrait
+
+
+@dataclass(init=False, frozen=True)
+class RestJson1Trait(Trait, id=ShapeID("aws.protocols#restJson1")):
+    http: Sequence[str] = field(
+        repr=False, hash=False, compare=False, default_factory=tuple
+    )
+    event_stream_http: Sequence[str] = field(
+        repr=False, hash=False, compare=False, default_factory=tuple
+    )
+
+    def __init__(self, value: DocumentValue | DynamicTrait = None):
+        super().__init__(value)
+        assert isinstance(self.document_value, Mapping)
+
+        http_versions = self.document_value["http"]
+        assert isinstance(http_versions, Sequence)
+        for val in http_versions:
+            assert isinstance(val, str)
+        object.__setattr__(self, "http", tuple(http_versions))
+        event_stream_http_versions = self.document_value.get("eventStreamHttp")
+        if not event_stream_http_versions:
+            object.__setattr__(self, "event_stream_http", self.http)
+        else:
+            assert isinstance(event_stream_http_versions, Sequence)
+            for val in event_stream_http_versions:
+                assert isinstance(val, str)
+            object.__setattr__(
+                self, "event_stream_http", tuple(event_stream_http_versions)
+            )

packages/smithy-http/src/smithy_http/aio/protocols.py

@@ -0,0 +1,117 @@
+import os
+from inspect import iscoroutinefunction
+from io import BytesIO
+
+from smithy_core.aio.interfaces import ClientProtocol
+from smithy_core.codecs import Codec
+from smithy_core.deserializers import DeserializeableShape
+from smithy_core.documents import TypeRegistry
+from smithy_core.exceptions import ExpectationNotMetException
+from smithy_core.interfaces import Endpoint, TypedProperties, URI
+from smithy_core.schemas import APIOperation
+from smithy_core.serializers import SerializeableShape
+from smithy_core.traits import HTTPTrait, EndpointTrait
+from smithy_http.aio.interfaces import HTTPRequest, HTTPResponse
+from smithy_http.deserializers import HTTPResponseDeserializer
+from smithy_http.serializers import HTTPRequestSerializer
+
+
+class HttpClientProtocol(ClientProtocol[HTTPRequest, HTTPResponse]):
+    """An HTTP-based protocol."""
+
+    def set_service_endpoint(
+        self,
+        *,
+        request: HTTPRequest,
+        endpoint: Endpoint,
+    ) -> HTTPRequest:
+        uri = endpoint.uri
+        uri_builder = request.destination
+
+        if uri.scheme:
+            uri_builder.scheme = uri.scheme
+        if uri.host:
+            uri_builder.host = uri.host
+        if uri.port and uri.port > -1:
+            uri_builder.port = uri.port
+        if uri.path:
+            uri_builder.path = os.path.join(uri.path, uri_builder.path or "")
+        # TODO: merge headers from the endpoint properties bag
+        return request
+
+
+class HttpBindingClientProtocol(HttpClientProtocol):
+    """An HTTP-based protocol that uses HTTP binding traits."""
+
+    @property
+    def payload_codec(self) -> Codec:
+        """The codec used for the serde of input and output payloads."""
+        ...
+
+    @property
+    def content_type(self) -> str:
+        """The media type of the http payload."""
+        ...
+
+    def serialize_request[
+        OperationInput: "SerializeableShape",
+        OperationOutput: "DeserializeableShape",
+    ](
+        self,
+        *,
+        operation: APIOperation[OperationInput, OperationOutput],
+        input: OperationInput,
+        endpoint: URI,
+        context: TypedProperties,
+    ) -> HTTPRequest:
+        # TODO(optimization): request binding cache like done in SJ
+        serializer = HTTPRequestSerializer(
+            payload_codec=self.payload_codec,
+            http_trait=operation.schema.expect_trait(HTTPTrait),
+            endpoint_trait=operation.schema.get_trait(EndpointTrait),
+        )
+
+        input.serialize(serializer=serializer)
+        request = serializer.result
+
+        if request is None:
+            raise ExpectationNotMetException(
+                "Expected request to be serialized, but was None"
+            )
+
+        return request
+
+    async def deserialize_response[
+        OperationInput: "SerializeableShape",
+        OperationOutput: "DeserializeableShape",
+    ](
+        self,
+        *,
+        operation: APIOperation[OperationInput, OperationOutput],
+        request: HTTPRequest,
+        response: HTTPResponse,
+        error_registry: TypeRegistry,
+        context: TypedProperties,
+    ) -> OperationOutput:
+        if not (200 <= response.status <= 299):
+            # TODO: implement error serde from type registry
+            raise NotImplementedError
+
+        body = response.body
+
+        # if body is not streaming and is async, we have to buffer it
+        if not operation.output_stream_member:
+            if (
+                read := getattr(body, "read", None)
+            ) is not None and iscoroutinefunction(read):
+                body = BytesIO(await read())
+
+        # TODO(optimization): response binding cache like done in SJ
+        deserializer = HTTPResponseDeserializer(
+            payload_codec=self.payload_codec,
+            http_trait=operation.schema.expect_trait(HTTPTrait),
+            response=response,
+            body=body,  # type: ignore
+        )
+
+        return operation.output.deserialize(deserializer)

packages/smithy-http/src/smithy_http/serializers.py

@@ -29,6 +29,8 @@
     TimestampFormatTrait,
     EndpointTrait,
     HTTPErrorTrait,
+    MediaTypeTrait,
+    StreamingTrait,
 )
 from smithy_core.shapes import ShapeType
 from smithy_core.utils import serialize_float
@@ -83,22 +85,33 @@ def begin_struct(self, schema: Schema) -> Iterator[ShapeSerializer]:
         if self._endpoint_trait is not None:
             host_prefix = self._endpoint_trait.host_prefix
 
+        content_type = self._payload_codec.media_type
+
         if (payload_member := self._get_payload_member(schema)) is not None:
             if payload_member.shape_type in (ShapeType.BLOB, ShapeType.STRING):
+                content_type = (
+                    "application/octet-stream"
+                    if payload_member.shape_type is ShapeType.BLOB
+                    else "text/plain"
+                )
                 payload_serializer = RawPayloadSerializer()
                 binding_serializer = HTTPRequestBindingSerializer(
                     payload_serializer, self._http_trait.path, host_prefix
                 )
                 yield binding_serializer
                 payload = payload_serializer.payload
             else:
+                if (media_type := payload_member.get_trait(MediaTypeTrait)) is not None:
+                    content_type = media_type.value
                 payload = BytesIO()
                 payload_serializer = self._payload_codec.create_serializer(payload)
                 binding_serializer = HTTPRequestBindingSerializer(
                     payload_serializer, self._http_trait.path, host_prefix
                 )
                 yield binding_serializer
         else:
+            if self._get_eventstreaming_member(schema) is not None:
+                content_type = "application/vnd.amazon.eventstream"
             payload = BytesIO()
             payload_serializer = self._payload_codec.create_serializer(payload)
             with payload_serializer.begin_struct(schema) as body_serializer:
@@ -112,6 +125,10 @@ def begin_struct(self, schema: Schema) -> Iterator[ShapeSerializer]:
         ) is not None and not iscoroutinefunction(seek):
             seek(0)
 
+        # TODO: conditional on empty-ness and based on the protocol
+        headers = binding_serializer.header_serializer.headers
+        headers.append(("content-type", content_type))
+
         self.result = _HTTPRequest(
             method=self._http_trait.method,
             destination=URI(
@@ -122,7 +139,7 @@ def begin_struct(self, schema: Schema) -> Iterator[ShapeSerializer]:
                     prefix=self._http_trait.query or "",
                 ),
             ),
-            fields=tuples_to_fields(binding_serializer.header_serializer.headers),
+            fields=tuples_to_fields(headers),
             body=payload,
         )
 
@@ -132,6 +149,15 @@ def _get_payload_member(self, schema: Schema) -> Schema | None:
                 return member
         return None
 
+    def _get_eventstreaming_member(self, schema: Schema) -> Schema | None:
+        for member in schema.members.values():
+            if (
+                member.get_trait(StreamingTrait) is not None
+                and member.shape_type is ShapeType.UNION
+            ):
+                return member
+        return None
+
 
 class HTTPRequestBindingSerializer(InterceptingSerializer):
     """Delegates HTTP request bindings to binding-location-specific serializers."""

packages/smithy-http/tests/unit/test_serializers.py

@@ -44,7 +44,7 @@
 from smithy_http.deserializers import HTTPResponseDeserializer
 from smithy_json import JSONCodec
 from smithy_http.aio import HTTPResponse as _HTTPResponse
-from smithy_http import tuples_to_fields, Fields
+from smithy_http import tuples_to_fields, Field, Fields
 from smithy_http.serializers import HTTPRequestSerializer, HTTPResponseSerializer
 
 # TODO: empty header prefix, query map
@@ -1572,11 +1572,16 @@ def payload_cases() -> list[HTTPMessageTestCase]:
         ),
         HTTPMessageTestCase(
             HTTPStringPayload(payload="foo"),
-            HTTPMessage(body=b"foo"),
+            HTTPMessage(
+                fields=tuples_to_fields([("content-type", "text/plain")]), body=b"foo"
+            ),
         ),
         HTTPMessageTestCase(
             HTTPBlobPayload(payload=b"\xde\xad\xbe\xef"),
-            HTTPMessage(body=b"\xde\xad\xbe\xef"),
+            HTTPMessage(
+                fields=tuples_to_fields([("content-type", "application/octet-stream")]),
+                body=b"\xde\xad\xbe\xef",
+            ),
         ),
         HTTPMessageTestCase(
             HTTPStructuredPayload(payload=HTTPStringPayload(payload="foo")),
@@ -1589,7 +1594,10 @@ def async_streaming_payload_cases() -> list[HTTPMessageTestCase]:
     return [
         HTTPMessageTestCase(
             HTTPStreamingPayload(payload=AsyncBytesReader(b"\xde\xad\xbe\xef")),
-            HTTPMessage(body=AsyncBytesReader(b"\xde\xad\xbe\xef")),
+            HTTPMessage(
+                fields=tuples_to_fields([("content-type", "application/octet-stream")]),
+                body=AsyncBytesReader(b"\xde\xad\xbe\xef"),
+            ),
         ),
     ]
 
@@ -1604,6 +1612,8 @@ def async_streaming_payload_cases() -> list[HTTPMessageTestCase]:
     + async_streaming_payload_cases()
 )
 
+CONTENT_TYPE_FIELD = Field(name="content-type", values=["application/json"])
+
 
 @pytest.mark.parametrize("case", REQUEST_SER_CASES)
 async def test_serialize_http_request(case: HTTPMessageTestCase) -> None:
@@ -1623,6 +1633,10 @@ async def test_serialize_http_request(case: HTTPMessageTestCase) -> None:
     actual_query = actual.destination.query or ""
     expected_query = case.request.destination.query or ""
     assert actual_query == expected_query
+    # set the content-type field here, otherwise cases would have to duplicate it everywhere,
+    # but if the field is already set in the case, don't override it
+    if expected.fields.get(CONTENT_TYPE_FIELD.name) is None:
+        expected.fields.set_field(CONTENT_TYPE_FIELD)
     assert actual.fields == expected.fields
 
     if case.request.body:
@@ -1647,6 +1661,9 @@ async def test_serialize_http_response(case: HTTPMessageTestCase) -> None:
     expected = case.request
 
     assert actual is not None
+    # Remove content-type from expected, we're re-using the request cases for brevity
+    if expected.fields.get(CONTENT_TYPE_FIELD.name) is not None:
+        del expected.fields[CONTENT_TYPE_FIELD.name]
     assert actual.fields == expected.fields
     assert actual.status == expected.status