Fix custom type type casting used in multiple attributes

pydantic/_internal/_generate_schema.py

@@ -52,6 +52,7 @@
     define_expected_missing_refs,
     get_ref,
     get_type_ref,
+    is_function_with_inner_schema,
     is_list_like_schema_with_items_schema,
     simplify_schema_references,
     validate_core_schema,
@@ -619,7 +620,13 @@ def _generate_schema_from_property(self, obj: Any, source: Any) -> core_schema.C
 
         schema = self._unpack_refs_defs(schema)
 
-        ref = get_ref(schema)
+        if is_function_with_inner_schema(schema):
+            ref = schema['schema'].pop('ref', None)
+            if ref:
+                schema['ref'] = ref
+        else:
+            ref = get_ref(schema)
+
         if ref:
             self.defs.definitions[ref] = self._post_process_generated_schema(schema)
             return core_schema.definition_reference_schema(ref)

tests/test_json_schema.py

@@ -33,10 +33,11 @@
 import pytest
 from dirty_equals import HasRepr
 from pydantic_core import CoreSchema, SchemaValidator, core_schema, to_json
-from typing_extensions import Annotated, Literal, TypedDict
+from typing_extensions import Annotated, Literal, Self, TypedDict
 
 import pydantic
 from pydantic import (
+    AfterValidator,
     BaseModel,
     Field,
     GetCoreSchemaHandler,
@@ -5779,3 +5780,35 @@ class Foo(BaseModel):
             }
         }
     }
+
+
+def test_repeated_custom_type():
+    class Numeric(pydantic.BaseModel):
+        value: float
+
+        @classmethod
+        def __get_pydantic_core_schema__(cls, source_type: Any, handler: pydantic.GetCoreSchemaHandler) -> CoreSchema:
+            return core_schema.no_info_before_validator_function(cls.validate, handler(source_type))
+
+        @classmethod
+        def validate(cls, v: Any) -> Union[Dict[str, Any], Self]:
+            if isinstance(v, (str, float, int)):
+                return cls(value=v)
+            if isinstance(v, Numeric):
+                return v
+            if isinstance(v, dict):
+                return v
+            raise ValueError(f'Invalid value for {cls}: {v}')
+
+    def is_positive(value: Numeric):
+        assert value.value > 0.0, 'Must be positive'
+
+    class OuterModel(pydantic.BaseModel):
+        x: Numeric
+        y: Numeric
+        z: Annotated[Numeric, AfterValidator(is_positive)]
+
+    assert OuterModel(x=2, y=-1, z=1)
+
+    with pytest.raises(ValidationError):
+        OuterModel(x=2, y=-1, z=-1)