Make parse classmethod (#96)

* Make parse a class method

* Fix tests

* Fix typechecking

Author: AdrienVannson

src/betterproto2/__init__.py

@@ -768,14 +768,9 @@ def SerializeToString(self) -> bytes:
         """
         return bytes(self)
 
-    def __getstate__(self) -> bytes:
-        return bytes(self)
-
-    def __setstate__(self: T, pickled_bytes: bytes) -> T:
-        return self.parse(pickled_bytes)
-
     def __reduce__(self) -> tuple[Any, ...]:
-        return (self.__class__.FromString, (bytes(self),))
+        # To support pickling
+        return (self.__class__.parse, (bytes(self),))
 
     @classmethod
     def _type_hint(cls, field_name: str) -> type:
@@ -829,12 +824,12 @@ def _postprocess_single(self, wire_type: int, meta: FieldMetadata, field_name: s
                 else:
                     msg_cls = self._betterproto.cls_by_field[field_name]
 
-                value = msg_cls().parse(value)
+                value = msg_cls.parse(value)
 
                 if meta.unwrap:
                     value = value.to_wrapped()
             elif meta.proto_type == TYPE_MAP:
-                value = self._betterproto.cls_by_field[field_name]().parse(value)
+                value = self._betterproto.cls_by_field[field_name].parse(value)
 
         return value
 
@@ -931,7 +926,8 @@ def load(
 
         return self
 
-    def parse(self: T, data: bytes) -> T:
+    @classmethod
+    def parse(cls, data: bytes) -> Self:
         """
         Parse the binary encoded Protobuf into this message instance. This
         returns the instance itself and is therefore assignable and chainable.
@@ -947,7 +943,7 @@ def parse(self: T, data: bytes) -> T:
             The initialized message.
         """
         with BytesIO(data) as stream:
-            return self.load(stream)
+            return cls().load(stream)
 
     # For compatibility with other libraries.
     @classmethod
@@ -971,7 +967,7 @@ def FromString(cls: type[T], data: bytes) -> T:
         :class:`Message`
             The initialized message.
         """
-        return cls().parse(data)
+        return cls.parse(data)
 
     def to_dict(
         self,
@@ -1210,21 +1206,22 @@ def _validate_field_groups(cls, values):
 Message.__annotations__ = {}  # HACK to avoid typing.get_type_hints breaking :)
 
 
-try:
-    import betterproto2_rust_codec  # pyright: ignore[reportMissingImports]
-except ModuleNotFoundError:
-    pass
-else:
+# The Rust codec is not available for now
+# try:
+#     import betterproto2_rust_codec  # pyright: ignore[reportMissingImports]
+# except ModuleNotFoundError:
+#     pass
+# else:
 
-    def parse_patched(self, data: bytes) -> Message:
-        betterproto2_rust_codec.deserialize(self, data)
-        return self
+#     def parse_patched(self, data: bytes) -> Message:
+#         betterproto2_rust_codec.deserialize(self, data)
+#         return self
 
-    def bytes_patched(self) -> bytes:
-        return betterproto2_rust_codec.serialize(self)
+#     def bytes_patched(self) -> bytes:
+#         return betterproto2_rust_codec.serialize(self)
 
-    Message.parse = parse_patched
-    Message.__bytes__ = bytes_patched
+#     Message.parse = parse_patched
+#     Message.__bytes__ = bytes_patched
 
 
 def which_one_of(message: Message, group_name: str) -> tuple[str, Any | None]:

tests/inputs/enum/test_enum.py

@@ -75,17 +75,17 @@ def enum_generator():
 
 def test_enum_mapped_on_parse():
     # test default value
-    b = Test().parse(bytes(Test()))
+    b = Test.parse(bytes(Test()))
     assert b.choice.name == Choice.ZERO.name
     assert b.choices == []
 
     # test non default value
-    a = Test().parse(bytes(Test(choice=Choice.ONE)))
+    a = Test.parse(bytes(Test(choice=Choice.ONE)))
     assert a.choice.name == Choice.ONE.name
     assert b.choices == []
 
     # test repeated
-    c = Test().parse(bytes(Test(choices=[Choice.THREE, Choice.FOUR])))
+    c = Test.parse(bytes(Test(choices=[Choice.THREE, Choice.FOUR])))
     assert c.choices[0].name == Choice.THREE.name
     assert c.choices[1].name == Choice.FOUR.name
 

tests/inputs/google_impl_behavior_equivalence/test_google_impl_behavior_equivalence.py

@@ -48,7 +48,7 @@ def test_bytes_are_the_same_for_oneof():
 
     assert message_bytes == message_reference_bytes
 
-    message2 = Test().parse(message_reference_bytes)
+    message2 = Test.parse(message_reference_bytes)
     message_reference2 = ReferenceTest()
     message_reference2.ParseFromString(message_reference_bytes)
 
@@ -71,7 +71,7 @@ def test_datetime_clamping(dt):  # see #407
     assert bytes(Spam(dt)) == ReferenceSpam(ts=ts).SerializeToString()
     message_bytes = bytes(Spam(dt))
 
-    assert Spam().parse(message_bytes).ts.timestamp() == ReferenceSpam.FromString(message_bytes).ts.seconds
+    assert Spam.parse(message_bytes).ts.timestamp() == ReferenceSpam.FromString(message_bytes).ts.seconds
 
 
 def test_empty_message_field():

tests/inputs/regression_387/test_regression_387.py

@@ -1,12 +1,9 @@
-from tests.output_betterproto.regression_387 import (
-    ParentElement,
-    Test,
-)
+from tests.output_betterproto.regression_387 import ParentElement, Test
 
 
 def test_regression_387():
     el = ParentElement(name="test", elems=[Test(id=0), Test(id=42)])
     binary = bytes(el)
-    decoded = ParentElement().parse(binary)
+    decoded = ParentElement.parse(binary)
     assert decoded == el
     assert decoded.elems == [Test(id=0), Test(id=42)]

tests/inputs/regression_414/test_regression_414.py

@@ -8,7 +8,7 @@ def test_full_cycle():
 
     obj = Test(body=body, auth=auth, signatures=sig)
 
-    decoded = Test().parse(bytes(obj))
+    decoded = Test.parse(bytes(obj))
     assert decoded == obj
     assert decoded.body == body
     assert decoded.auth == auth

tests/test_any.py

@@ -8,7 +8,7 @@ def test_any() -> None:
     any = Any()
     any.pack(person)
 
-    new_any = Any().parse(bytes(any))
+    new_any = Any.parse(bytes(any))
 
     assert new_any.unpack() == person
 

tests/test_features.py

@@ -47,10 +47,10 @@ def test_unknown_fields():
     serialized_newer = bytes(newer)
 
     # Unknown fields in `Newer` should round trip with `Older`
-    round_trip = bytes(Older().parse(serialized_newer))
+    round_trip = bytes(Older.parse(serialized_newer))
     assert serialized_newer == round_trip
 
-    new_again = Newer().parse(round_trip)
+    new_again = Newer.parse(round_trip)
     assert newer == new_again
 
 
@@ -84,7 +84,7 @@ def test_oneof_support():
     assert bytes(msg) == b"\x08\x00"
 
     # Round trip should also work
-    msg = OneofMsg().parse(bytes(msg))
+    msg = OneofMsg.parse(bytes(msg))
     assert betterproto2.which_one_of(msg, "group1")[0] == "x"
     assert msg.x == 0
     assert betterproto2.which_one_of(msg, "group2")[0] == ""
@@ -147,8 +147,8 @@ def test_optional_flag():
     assert bytes(OptionalBoolMsg(field=False)) == b"\n\x00"
 
     # Differentiate between not passed and the zero-value.
-    assert OptionalBoolMsg().parse(b"").field is None
-    assert OptionalBoolMsg().parse(b"\n\x00").field is False
+    assert OptionalBoolMsg.parse(b"").field is None
+    assert OptionalBoolMsg.parse(b"\n\x00").field is False
 
 
 def test_optional_datetime_to_dict():
@@ -271,7 +271,7 @@ def test_oneof_default_value_set_causes_writes_wire():
     from tests.output_betterproto.features import Empty, MsgC
 
     def _round_trip_serialization(msg: MsgC) -> MsgC:
-        return MsgC().parse(bytes(msg))
+        return MsgC.parse(bytes(msg))
 
     int_msg = MsgC(int_field=0)
     str_msg = MsgC(string_field="")

tests/test_pickling.py

@@ -1,8 +1,5 @@
 import pickle
-from copy import (
-    copy,
-    deepcopy,
-)
+from copy import copy, deepcopy
 
 import cachelib
 
@@ -49,10 +46,7 @@ def test_pickling_complex_message():
 
 
 def test_recursive_message_defaults():
-    from tests.output_betterproto.recursivemessage import (
-        Intermediate,
-        Test as RecursiveMessage,
-    )
+    from tests.output_betterproto.recursivemessage import Intermediate, Test as RecursiveMessage
 
     msg = RecursiveMessage(name="bob", intermediate=Intermediate(42))
     msg = unpickled(msg)