only support partial on validators that take ctx

Author: petersli

docs/output.md

@@ -472,9 +472,7 @@ _(This example is complete, it can be run "as is")_
 
 #### Partial validation during streaming {#partial-validation}
 
-When [streaming responses](#streaming-model-responses), you can pass `allow_partial=True` to `validate_response_output` to make Pydantic output validators use partial validation.
-
-By adding a `partial: bool` parameter as the last argument, you can access this information within output validator functions as well.
+If your output validator takes `RunContext` as a first argument, you can add `partial: bool` as the last argument to handle validation differently for partial outputs during streaming (e.g. skip validation altogether). 
 
 ```python
 from pydantic_ai import Agent, ModelRetry, RunContext
@@ -486,8 +484,8 @@ def validate_output(ctx: RunContext, output: str, partial: bool) -> str:
     if partial:
         return output
     else:
-        if 'invalid' in output:
-            raise ModelRetry('Output contains invalid content')
+        if len(output) < 100:
+            raise ModelRetry('Output is too short.')
         return output
 ```
 

pydantic_ai_slim/pydantic_ai/_output.py

@@ -60,15 +60,13 @@
     | Callable[[RunContext[AgentDepsT], OutputDataT_inv], Awaitable[OutputDataT_inv]]
     | Callable[[OutputDataT_inv], OutputDataT_inv]
     | Callable[[OutputDataT_inv], Awaitable[OutputDataT_inv]]
-    | Callable[[OutputDataT_inv, bool], OutputDataT_inv]
-    | Callable[[OutputDataT_inv, bool], Awaitable[OutputDataT_inv]]
 )
 """
-A function that always takes and returns the same type of data (which is the result type of an agent run), and:
+A function that always takes and returns the same type of data (which is the result type of an agent run). In addition:
 
-* may or may not take [`RunContext`][pydantic_ai.tools.RunContext] as a first argument
-* may or may not take a `partial: bool` parameter as a last argument
-* may or may not be async
+* it can optionally take [`RunContext`][pydantic_ai.tools.RunContext] as a first argument
+   * if it takes [`RunContext`][pydantic_ai.tools.RunContext] as a first argument, it can also optionally take `partial: bool` as a last argument
+* it can be async
 
 Usage `OutputValidatorFunc[AgentDepsT, T]`.
 """
@@ -171,15 +169,9 @@ class OutputValidator(Generic[AgentDepsT, OutputDataT_inv]):
     _is_async: bool = field(init=False)
 
     def __post_init__(self):
-        params = list(inspect.signature(self.function).parameters.values())
-        if params:
-            first_param = params[0]
-            annotation = first_param.annotation
-            self._takes_ctx = 'RunContext' in str(annotation)
-
-        expected_partial_index = 2 if self._takes_ctx else 1
-        self._takes_partial = len(params) > expected_partial_index
-
+        sig = inspect.signature(self.function)
+        self._takes_ctx = len(sig.parameters) > 1
+        self._takes_partial = len(sig.parameters) > 2
         self._is_async = _utils.is_async_callable(self.function)
 
     async def validate(