Backtestspydanticrun-20261007-130130#13460

Issue #13460 Merged fix PR #13487 Run 7 Oct 2026 · run-20261007-130130 pydantic/pydantic

Unexpected serialization schema with ValidateAs

Worse

The merged fix was judged better than our patch.

In this judgment, Patch B = our agent, Patch A = the merged fix.

Both patches fix the root cause the same way: they generate a core schema for the annotated `source` type and attach it as the serialization schema of the after-validator. This stops serialization from following `from_type`, which caused the `list[int]` warning on an `int` value.

Patch A is simpler and more robust. It always wraps the source schema in a `wrap_serializer_function_ser_schema` with `h(v)`. That still honors any custom serializer on the source schema, such as `PlainSerializer`, because the inner handler serializes through that schema. It falls back to `any_schema()` on the full `GENERATE_SCHEMA_ERRORS` tuple, matching the existing convention a few lines above in the same file (the `InstanceOf`/`SkipValidation` code).

Patch B is riskier. It reuses `source_schema['serialization']` directly when present. Its own TODO admits this can produce an invalid schema when the serialization is an include-exclude dict/sequence schema, which it papers over with type-ignore comments. It catches only `PydanticSchemaGenerationError`, which is narrower than the codebase's established error set. It also calls `handler.generate_schema(source)` a second time just to supply a `return_schema`, which is redundant work.

On tests, B is more thorough: it turns warnings into errors, checks both JSON schema modes, and covers a custom serializer on the source type and an arbitrary source type. A's tests are thinner but still meaningful: the JSON dump of the issue scenario, plus the unknown-schema fallback. B's better tests do not outweigh A's cleaner and less fragile implementation, which also avoids a known invalid-schema edge case.

The judge saw only the two patches, labelled A and B in random order, with issue links and version numbers removed. The reasoning above is shown verbatim.

Solve

11m 17s

Wall-clock time in the container

Full test suite

No new failures

12,302 passed · 501 skipped · 12,803 total

Contamination audit

Clean

No sign the agent saw the real fix

Patch size

80vs 56 merged

Changed lines, ours vs merged

Initial Checks

  • I confirm that I'm using Pydantic V2

Description

From https://github.com/pydantic/pydantic/discussions/13232:

import pydantic as p
import typing as t

def first_even(nums: list[int]) -> int:
    for i in nums:
        if i%2 == 0:
            return i
    raise ValueError("Even not found.")

class Foo(p.BaseModel):
    x: t.Annotated[int, p.ValidateAs(list[int], first_even)]

d = Foo(x=[1, 3, 5, 8, 6, 7])
print(d)
# x=8
print(d.model_dump())
# PydanticSerializationUnexpectedValue(Expected `list[int]` - serialized value may not be as expected [field_name='x', input_value=8, input_type=int])
# {'x': 8}

Example Code

Python, Pydantic & OS Version

main

The two patches

Our patch+80 −0
pydantic/functional_validators.pyindex 6a38f619..5e331f2d 100644@@ -894,8 +894,29 @@ class ValidateAs:         self.instantiation_hook = instantiation_hook      def __get_pydantic_core_schema__(self, source: Any, handler: GetCoreSchemaHandler) -> core_schema.CoreSchema:+        # The validated value is an instance of the annotated (source) type, not of `from_type`,+        # so serialization must follow the source type's schema. If Pydantic can't generate a core schema+        # for the source type (e.g. an unsupported custom type), the default serialization behavior is kept.+        from pydantic import PydanticSchemaGenerationError++        try:+            source_schema = handler(source)+            # TODO if `source_schema['serialization']` is one of `'include-exclude-dict/sequence',+            # schema validation will fail. That's why we use 'type ignore' comments below.+            serialization = source_schema.get(+                'serialization',+                core_schema.wrap_serializer_function_ser_schema(+                    function=lambda v, h: h(v),+                    schema=source_schema,+                    return_schema=handler.generate_schema(source),+                ),+            )+        except PydanticSchemaGenerationError:+            serialization = None+         schema = handler(self.from_type)         return core_schema.no_info_after_validator_function(             self.instantiation_hook,             schema=schema,+            serialization=serialization,  # pyright: ignore[reportArgumentType]         )tests/test_types.pyindex e70a0102..5a4a13e1 100644@@ -6096,6 +6096,65 @@ def test_validate_from() -> None:     assert ta.validate_python({'a': 1}) == Arbitrary(1)  +def test_validate_as_serializes_as_source_type() -> None:+    """https://github.com/pydantic/pydantic/discussions/13232"""++    def first_even(nums: list[int]) -> int:+        for i in nums:+            if i % 2 == 0:+                return i+        raise ValueError('Even not found.')++    class Foo(BaseModel):+        x: Annotated[int, ValidateAs(list[int], first_even)]++    with warnings.catch_warnings():+        warnings.simplefilter('error')+        foo = Foo(x=[1, 3, 5, 8, 6, 7])+        assert foo.x == 8+        assert foo.model_dump() == {'x': 8}+        assert foo.model_dump_json() == '{"x":8}'++    assert Foo.model_json_schema(mode='validation')['properties']['x'] == {+        'items': {'type': 'integer'},+        'title': 'X',+        'type': 'array',+    }+    assert Foo.model_json_schema(mode='serialization')['properties']['x'] == {'title': 'X', 'type': 'integer'}+++def test_validate_as_uses_source_type_custom_serializer() -> None:+    class Foo(BaseModel):+        x: Annotated[+            Annotated[int, PlainSerializer(lambda v: f'#{v}')],+            ValidateAs(str, int),+        ]++    with warnings.catch_warnings():+        warnings.simplefilter('error')+        foo = Foo(x='12')+        assert foo.x == 12+        assert foo.model_dump() == {'x': '#12'}+++def test_validate_as_arbitrary_source_type_serialization() -> None:+    class Arbitrary:+        def __init__(self, a: int) -> None:+            self.a = a++    class ArbitraryModel(BaseModel):+        a: int++    class Foo(BaseModel, arbitrary_types_allowed=True):+        x: Annotated[Arbitrary, ValidateAs(ArbitraryModel, instantiation_hook=lambda v: Arbitrary(a=v.a))]++    with warnings.catch_warnings():+        warnings.simplefilter('error')+        foo = Foo(x={'a': 1})+        assert isinstance(foo.x, Arbitrary)+        assert foo.model_dump()['x'] is foo.x++ @pytest.mark.skipif(sys.version_info < (3, 12), reason="`Annotated` doesn't allow instances in <3.12") def test_skip_validation_arbitrary_type_object() -> None:     """https://github.com/pydantic/pydantic/issues/11997. 
Merged fix · PR #13487+52 −4
pydantic/functional_validators.pyindex 6a38f6195..336eff24d 100644@@ -794,6 +794,7 @@ else:          @classmethod         def __get_pydantic_core_schema__(cls, source: Any, handler: GetCoreSchemaHandler) -> core_schema.CoreSchema:+            # Lazy import to avoid cycle:             from pydantic._internal._generate_schema import GENERATE_SCHEMA_ERRORS              # use the generic _origin_ as the second argument to isinstance when appropriate@@ -894,8 +895,24 @@ class ValidateAs:         self.instantiation_hook = instantiation_hook      def __get_pydantic_core_schema__(self, source: Any, handler: GetCoreSchemaHandler) -> core_schema.CoreSchema:-        schema = handler(self.from_type)-        return core_schema.no_info_after_validator_function(+        # Lazy import to avoid cycle:+        from pydantic._internal._generate_schema import GENERATE_SCHEMA_ERRORS++        from_type_schema = handler(self.from_type)+        schema = core_schema.no_info_after_validator_function(             self.instantiation_hook,-            schema=schema,+            schema=from_type_schema,+        )+        try:+            ser_schema = handler(source)+        except GENERATE_SCHEMA_ERRORS:+            ser_schema = core_schema.any_schema()++        # TODO: pydantic-core currently doesn't accept it, but we should be able to provide an+        # arbitrary core schema for serialization (currently `simple_ser_schema()` can be used,+        # but only allow specifying a specific `'type`). Working around this with a wrap serializer+        # is already done in other parts of the code base, so grep for similar cases when fixing:+        schema['serialization'] = core_schema.wrap_serializer_function_ser_schema(+            function=lambda v, h: h(v), schema=ser_schema         )+        return schematests/test_types.pyindex e70a0102e..df4d5f060 100644@@ -6080,7 +6080,7 @@ def test_skip_validation_json_schema():     }  -def test_validate_from() -> None:+def test_validate_as() -> None:     class Arbitrary:         def __init__(self, a: int) -> None:             self.a = a@@ -6096,6 +6096,37 @@ def test_validate_from() -> None:     assert ta.validate_python({'a': 1}) == Arbitrary(1)  +def test_validate_as_known_serialization() -> None:+    """https://github.com/pydantic/pydantic/issues/13460"""++    def first_even(nums: list[int]) -> int:+        for i in nums:+            if i % 2 == 0:+                return i+        raise ValueError('Even not found.')++    class Model(BaseModel):+        x: Annotated[int, ValidateAs(list[int], first_even)]++    m = Model(x=[1, 2])++    assert m.model_dump_json() == '{"x":2}'+++def test_validate_as_unknown_serialization() -> None:++    class Arbitrary:+        def __init__(self, a: int) -> None:+            self.a = a++        def __eq__(self, other) -> bool:+            return self.a == other.a++    ta = TypeAdapter(Annotated[Arbitrary, ValidateAs(int, instantiation_hook=Arbitrary)])++    assert ta.dump_python(Arbitrary(1)) == Arbitrary(1)++ @pytest.mark.skipif(sys.version_info < (3, 12), reason="`Annotated` doesn't allow instances in <3.12") def test_skip_validation_arbitrary_type_object() -> None:     """https://github.com/pydantic/pydantic/issues/11997.