Draft: Protobuf SerDes support

pyproject.toml

@@ -23,6 +23,11 @@ classifiers = [
     "Programming Language :: Python :: 3",
 ]
 
+[project.optional-dependencies]
+protobuf = [
+  "protobuf"
+]
+
 [project.urls]
 Homepage = "https://github.com/quixio/quix-streams"
 

quixstreams/models/serializers/protobuf.py

@@ -0,0 +1,56 @@
+from typing import Union, Mapping, Iterable, Dict
+
+from .base import Serializer, Deserializer, SerializationContext
+from .exceptions import SerializationError
+
+from google.protobuf.message import Message, DecodeError, EncodeError
+from google.protobuf.json_format import MessageToDict, ParseDict, ParseError
+
+__all__ = ("ProtobufSerializer", "ProtobufDeserializer")
+
+
+class ProtobufSerializer(Serializer):
+    def __init__(
+        self,
+        msg_type: Message,
+    ):
+        """
+        Serializer that returns data in protobuf format.
+
+        :param msg_type: protobuf message class.
+        """
+        super().__init__()
+        self._msg_type = msg_type
+
+    def __call__(self, value: Dict, ctx: SerializationContext) -> Union[str, bytes]:
+        msg = self._msg_type()
+
+        try:
+            return ParseDict(value, msg).SerializeToString(deterministic=True)
+        except (EncodeError, ParseError) as exc:
+            raise SerializationError(str(exc)) from exc
+
+
+class ProtobufDeserializer(Deserializer):
+    def __init__(
+        self,
+        msg_type: Message,
+    ):
+        """
+        Deserializer that parses protobuf data.
+
+        :param msg_type: protobuf message class.
+        """
+        super().__init__()
+        self._msg_type = msg_type
+
+    def __call__(
+        self, value: bytes, ctx: SerializationContext
+    ) -> Union[Iterable[Mapping], Mapping]:
+        msg = self._msg_type()
+
+        try:
+            msg.ParseFromString(value)
+            return MessageToDict(msg)
+        except DecodeError as exc:
+            raise SerializationError(str(exc)) from exc

tests/test_quixstreams/test_models/protobuf/test.proto

@@ -0,0 +1,8 @@
+syntax = "proto3";
+
+package test;
+
+message Test {
+  string name = 1;
+  int32 id = 2;
+}

tests/test_quixstreams/test_models/test_serializers.py

@@ -16,8 +16,14 @@
     DoubleDeserializer,
     StringDeserializer,
 )
+from quixstreams.models.serializers.protobuf import (
+    ProtobufSerializer,
+    ProtobufDeserializer,
+)
 from .utils import int_to_bytes, float_to_bytes
 
+from .protobuf.test_pb2 import Test
+
 dummy_context = SerializationContext(topic="topic")
 
 
@@ -34,6 +40,10 @@ class TestSerializers:
             (BytesSerializer(), b"abc", b"abc"),
             (JSONSerializer(), {"a": 123}, b'{"a":123}'),
             (JSONSerializer(), [1, 2, 3], b"[1,2,3]"),
+            (ProtobufSerializer(Test), {}, b""),
+            (ProtobufSerializer(Test), {"id": 3}, b"\x10\x03"),
+            (ProtobufSerializer(Test), {"name": "foo", "id": 2}, b"\n\x03foo\x10\x02"),
+            (ProtobufSerializer(Test), {"name": "foo"}, b"\n\x03foo"),
         ],
     )
     def test_serialize_success(self, serializer: Serializer, value, expected):
@@ -50,6 +60,7 @@ def test_serialize_success(self, serializer: Serializer, value, expected):
             (StringSerializer(), {"a": 123}),
             (JSONSerializer(), object()),
             (JSONSerializer(), complex(1, 2)),
+            (ProtobufSerializer(Test), {"bar": 3}),
         ],
     )
     def test_serialize_error(self, serializer: Serializer, value):
@@ -70,6 +81,14 @@ class TestDeserializers:
             (BytesDeserializer(), b"123123", b"123123"),
             (JSONDeserializer(), b"123123", 123123),
             (JSONDeserializer(), b'{"a":"b"}', {"a": "b"}),
+            (
+                ProtobufDeserializer(Test),
+                b"\n\x03foo\x10\x02",
+                {"name": "foo", "id": 2},
+            ),
+            (ProtobufDeserializer(Test), b"\n\x03foo", {"name": "foo"}),
+            (ProtobufDeserializer(Test), b"\x10\x03", {"id": 3}),
+            (ProtobufDeserializer(Test), b"", {}),
         ],
     )
     def test_deserialize_no_column_name_success(
@@ -84,6 +103,7 @@ def test_deserialize_no_column_name_success(
             (IntegerDeserializer(), b'{"abc": "abc"}'),
             (DoubleDeserializer(), b"abc"),
             (JSONDeserializer(), b"{"),
+            (ProtobufDeserializer(Test), b"\n\x03foo\x10\x02\x13"),
         ],
     )
     def test_deserialize_error(self, deserializer: Deserializer, value):