Use external package betterproto-rust-codec for better (de-)serialization performance (#545)

* optionally use betterproto-rust-codec
* monkey patch `parse` and `__bytes__`

benchmarks/benchmarks.py

@@ -6,32 +6,32 @@ import betterproto
 
 @dataclass
 class TestMessage(betterproto.Message):
-    foo: int = betterproto.uint32_field(0)
+    foo: int = betterproto.uint32_field(1)
-    bar: str = betterproto.string_field(1)
+    bar: str = betterproto.string_field(2)
-    baz: float = betterproto.float_field(2)
+    baz: float = betterproto.float_field(3)
 
 
 @dataclass
 class TestNestedChildMessage(betterproto.Message):
-    str_key: str = betterproto.string_field(0)
+    str_key: str = betterproto.string_field(1)
-    bytes_key: bytes = betterproto.bytes_field(1)
+    bytes_key: bytes = betterproto.bytes_field(2)
-    bool_key: bool = betterproto.bool_field(2)
+    bool_key: bool = betterproto.bool_field(3)
-    float_key: float = betterproto.float_field(3)
+    float_key: float = betterproto.float_field(4)
-    int_key: int = betterproto.uint64_field(4)
+    int_key: int = betterproto.uint64_field(5)
 
 
 @dataclass
 class TestNestedMessage(betterproto.Message):
-    foo: TestNestedChildMessage = betterproto.message_field(0)
+    foo: TestNestedChildMessage = betterproto.message_field(1)
-    bar: TestNestedChildMessage = betterproto.message_field(1)
+    bar: TestNestedChildMessage = betterproto.message_field(2)
-    baz: TestNestedChildMessage = betterproto.message_field(2)
+    baz: TestNestedChildMessage = betterproto.message_field(3)
 
 
 @dataclass
 class TestRepeatedMessage(betterproto.Message):
-    foo_repeat: List[str] = betterproto.string_field(0)
+    foo_repeat: List[str] = betterproto.string_field(1)
-    bar_repeat: List[int] = betterproto.int64_field(1)
+    bar_repeat: List[int] = betterproto.int64_field(2)
-    baz_repeat: List[bool] = betterproto.bool_field(2)
+    baz_repeat: List[bool] = betterproto.bool_field(3)
 
 
 class BenchMessage:
@@ -44,25 +44,14 @@ class BenchMessage:
         self.instance_filled_bytes = bytes(self.instance_filled)
         self.instance_filled_nested = TestNestedMessage(
             TestNestedChildMessage("foo", bytearray(b"test1"), True, 0.1234, 500),
-            TestNestedChildMessage("bar", bytearray(b"test2"), True, 3.1415, -302),
+            TestNestedChildMessage("bar", bytearray(b"test2"), True, 3.1415, 302),
             TestNestedChildMessage("baz", bytearray(b"test3"), False, 1e5, 300),
         )
         self.instance_filled_nested_bytes = bytes(self.instance_filled_nested)
         self.instance_filled_repeated = TestRepeatedMessage(
-            [
+            [f"test{i}" for i in range(1_000)],
-                "test1",
+            [(i - 500) ** 3 for i in range(1_000)],
-                "test2",
+            [i % 2 == 0 for i in range(1_000)],
-                "test3",
-                "test4",
-                "test5",
-                "test6",
-                "test7",
-                "test8",
-                "test9",
-                "test10",
-            ],
-            [2, -100, 0, 500000, 600, -425678, 1000000000, -300, 1, -694214214466],
-            [True, False, False, False, True, True, False, True, False, False],
         )
         self.instance_filled_repeated_bytes = bytes(self.instance_filled_repeated)
 
@@ -71,9 +60,9 @@ class BenchMessage:
 
         @dataclass
         class Message(betterproto.Message):
-            foo: int = betterproto.uint32_field(0)
+            foo: int = betterproto.uint32_field(1)
-            bar: str = betterproto.string_field(1)
+            bar: str = betterproto.string_field(2)
-            baz: float = betterproto.float_field(2)
+            baz: float = betterproto.float_field(3)
 
     def time_instantiation(self):
         """Time instantiation"""

poetry.lock

@@ -97,6 +97,26 @@ files = [
     {file = "backports.cached_property-1.0.2-py3-none-any.whl", hash = "sha256:baeb28e1cd619a3c9ab8941431fe34e8490861fb998c6c4590693d50171db0cc"},
 ]
 
+[[package]]
+name = "betterproto-rust-codec"
+version = "0.1.0"
+description = "Fast conversion between betterproto messages and Protobuf wire format."
+optional = true
+python-versions = ">=3.7"
+files = [
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-macosx_10_7_x86_64.whl", hash = "sha256:3e7bea088f5f630cf123b56b1e1da7b35a22a295561825dd7c1446fa841876c3"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-macosx_11_0_arm64.whl", hash = "sha256:d1706da37a162d72de557ae3e285d04bb66cf129584d380439266e6e0d60c70a"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:07aeb3471e9ea3f142f79673b1057810a133c8285aa0a606c53a43a3e28c37da"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:4b905251be0eae0e7605ecf4a2d60f8c54f3a3c5e6d1d732a0060a18d69cf5e4"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:c82353cd2a53f0ced0546d94dd5b18026eed1231a7658e6d5876cc09b1cfdae8"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:5b881d5a54e8ac7a898fc7f27971e91a6596d344faf342a262cecfa5b1f0dc1a"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9fc2d2d11d62a91b75d94783518563673972ad2c99bee9fd16176d3cd4c15d76"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:1f9b99f776ce1da52f83715b926b61f2ad52fdf1e6d7dec0aea5fb471cd468b1"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-win32.whl", hash = "sha256:2f98155920ff409551208820224ac62564e8dd89d7caed86f8cd519fc7e05568"},
+    {file = "betterproto_rust_codec-0.1.0-cp37-abi3-win_amd64.whl", hash = "sha256:8e32f75cf59d5faaa5efb00c268fa22f47def0855c3395e5accccf296b2243be"},
+    {file = "betterproto_rust_codec-0.1.0.tar.gz", hash = "sha256:090c23accb23e8f167dabab6fab70d30ae20f587e677c2da4a88982c5fafc5d1"},
+]
+
 [[package]]
 name = "black"
 version = "23.3.0"
@@ -1904,8 +1924,9 @@ testing = ["big-O", "flake8 (<5)", "jaraco.functools", "jaraco.itertools", "more
 
 [extras]
 compiler = ["black", "isort", "jinja2"]
+rust-codec = ["betterproto-rust-codec"]
 
 [metadata]
 lock-version = "2.0"
 python-versions = "^3.7"
-content-hash = "a522aa8f9c81809b07db9458b493302642836cb300e87efaeed441a451fb0558"
+content-hash = "0e0d9fe748a60b22a7147e05c7940ffcad6facb7554cc1ad791e58f4a4c5c22d"

pyproject.toml

@@ -20,6 +20,7 @@ jinja2 = { version = ">=3.0.3", optional = true }
 python-dateutil = "^2.8"
 isort = {version = "^5.11.5", optional = true}
 typing-extensions = "^4.7.1"
+betterproto-rust-codec = { version = "0.1.0", optional = true }
 
 [tool.poetry.group.dev.dependencies]
 asv = "^0.4.2"
@@ -48,6 +49,7 @@ protoc-gen-python_betterproto = "betterproto.plugin:main"
 
 [tool.poetry.extras]
 compiler = ["black", "isort", "jinja2"]
+rust-codec = ["betterproto-rust-codec"]
 
 
 # Dev workflow tasks

src/betterproto/__init__.py

@@ -1868,6 +1868,23 @@ class Message(ABC):
 
 Message.__annotations__ = {}  # HACK to avoid typing.get_type_hints breaking :)
 
+# monkey patch (de-)serialization functions of class `Message`
+# with functions from `betterproto-rust-codec` if available
+try:
+    import betterproto_rust_codec
+
+    def __parse_patch(self: T, data: bytes) -> T:
+        betterproto_rust_codec.deserialize(self, data)
+        return self
+
+    def __bytes_patch(self) -> bytes:
+        return betterproto_rust_codec.serialize(self)
+
+    Message.parse = __parse_patch
+    Message.__bytes__ = __bytes_patch
+except ModuleNotFoundError:
+    pass
+
 
 def serialized_on_wire(message: Message) -> bool:
     """