fix dataclass annotated before validator called twice

pydantic/fields.py

@@ -419,11 +419,7 @@ def _from_dataclass_field(cls, dc_field: DataclassField[Any]) -> typing_extensio
 
         # use the `Field` function so in correct kwargs raise the correct `TypeError`
         dc_field_metadata = {k: v for k, v in dc_field.metadata.items() if k in _FIELD_ARG_NAMES}
-        field = Field(default=default, default_factory=default_factory, repr=dc_field.repr, **dc_field_metadata)
-
-        field.annotation, annotation_metadata = cls._extract_metadata(dc_field.type)
-        field.metadata += annotation_metadata
-        return field
+        return Field(default=default, default_factory=default_factory, repr=dc_field.repr, **dc_field_metadata)
 
     @classmethod
     def _extract_metadata(cls, annotation: type[Any] | None) -> tuple[type[Any] | None, list[Any]]:

tests/test_dataclasses.py

@@ -18,6 +18,7 @@
 import pydantic
 from pydantic import (
     BaseModel,
+    BeforeValidator,
     ConfigDict,
     FieldValidationInfo,
     GenerateSchema,
@@ -2466,3 +2467,23 @@ class Model:
         __pydantic_config__ = ConfigDict(schema_generator=LaxStrGenerator)
 
     assert Model(x=1).x == '1'
+
+
+@pytest.mark.parametrize('decorator1', **dataclass_decorators())
+def test_annotated_before_validator_called_once(decorator1):
+    count = 0
+
+    def convert(value: int) -> str:
+        nonlocal count
+        count += 1
+        return str(value)
+
+    IntToStr = Annotated[str, BeforeValidator(convert)]
+
+    @decorator1
+    class A:
+        a: IntToStr
+
+    assert count == 0
+    TypeAdapter(A).validate_python({'a': 123})
+    assert count == 1