misc changes: returns None, root as ., add

src/wf_authoring/nodes/__init__.py

@@ -13,7 +13,7 @@ from .callables import (
 from .inference import accepts_context, infer_models, is_basemodel_subclass
 from .decorator import node
 from .registry import build_async_registry, build_registry
-from .result import NodeReturn, Nothing, outcome
+from .result import NoOutput, NodeReturn, Nothing, outcome
 from .schema import schema_ref_for
 from .spec import NodeSpec
 
@@ -25,6 +25,7 @@ __all__ = [
     "ContextNodeCallable",
     "InputT",
     "NodeCallable",
+    "NoOutput",
     "NodeReturn",
     "NodeSpec",
     "Nothing",

src/wf_authoring/nodes/callables.py

@@ -22,7 +22,7 @@ class ContextNodeCallable(Protocol[InputT_contra, OutputT_co]):
         payload: InputT_contra,
         /,
         ctx: RuntimeContext,
-    ) -> NodeReturn[OutputT_co] | OutputT_co: ...
+    ) -> NodeReturn[OutputT_co] | OutputT_co | None: ...
 
 
 class PlainNodeCallable(Protocol[InputT_contra, OutputT_co]):
@@ -30,7 +30,7 @@ class PlainNodeCallable(Protocol[InputT_contra, OutputT_co]):
         self,
         payload: InputT_contra,
         /,
-    ) -> NodeReturn[OutputT_co] | OutputT_co: ...
+    ) -> NodeReturn[OutputT_co] | OutputT_co | None: ...
 
 
 NodeCallable = ContextNodeCallable[InputT, OutputT] | PlainNodeCallable[InputT, OutputT]
@@ -42,7 +42,7 @@ class AsyncContextNodeCallable(Protocol[InputT_contra, OutputT_co]):
         payload: InputT_contra,
         /,
         ctx: RuntimeContext,
-    ) -> Awaitable[NodeReturn[OutputT_co] | OutputT_co]: ...
+    ) -> Awaitable[NodeReturn[OutputT_co] | OutputT_co | None]: ...
 
 
 class AsyncPlainNodeCallable(Protocol[InputT_contra, OutputT_co]):
@@ -50,7 +50,7 @@ class AsyncPlainNodeCallable(Protocol[InputT_contra, OutputT_co]):
         self,
         payload: InputT_contra,
         /,
-    ) -> Awaitable[NodeReturn[OutputT_co] | OutputT_co]: ...
+    ) -> Awaitable[NodeReturn[OutputT_co] | OutputT_co | None]: ...
 
 
 AsyncNodeCallable = (

src/wf_authoring/nodes/decorator.py

@@ -6,7 +6,12 @@ from typing import Any, cast, overload
 
 from pydantic import BaseModel
 
-from .callables import AsyncNodeCallable, InputT, NodeCallable, OutputT
+from .callables import (
+    AsyncNodeCallable,
+    InputT,
+    NodeCallable,
+    OutputT,
+)
 from .inference import accepts_context, infer_models
 from .spec import NodeSpec
 

src/wf_authoring/nodes/inference.py

@@ -8,7 +8,7 @@ from pydantic import BaseModel
 
 from wf_core import RuntimeContext
 
-from .result import NodeReturn
+from .result import NodeReturn, Nothing
 
 
 def is_basemodel_subclass(value: object) -> bool:
@@ -47,10 +47,15 @@ def infer_models(fn: Callable[..., object]) -> tuple[type[BaseModel], type[BaseM
     return_type = hints.get("return")
     if return_type is None:
         raise TypeError("node function must declare a return annotation")
+    if return_type is type(None):
+        return cast(type[BaseModel], input_model), Nothing
 
     if is_basemodel_subclass(return_type):
         return cast(type[BaseModel], input_model), cast(type[BaseModel], return_type)
 
+    if return_type is NodeReturn:
+        return cast(type[BaseModel], input_model), Nothing
+
     origin = get_origin(return_type)
     if origin is NodeReturn:
         args = get_args(return_type)

src/wf_authoring/nodes/result.py

@@ -20,6 +20,10 @@ class Nothing(BaseModel):
     """Empty output model for nodes that only choose an outcome."""
 
 
+NoOutput = NodeReturn[Nothing]
+"""Type alias for outcome-only nodes that return no output payload."""
+
+
 @overload
 def outcome(name: str) -> NodeReturn[Nothing]: ...
 

src/wf_authoring/nodes/spec.py

@@ -18,7 +18,7 @@ from .callables import (
     PlainNodeCallable,
     SyncRegistryHandler,
 )
-from .result import NodeReturn
+from .result import NodeReturn, Nothing
 from .schema import schema_ref_for
 
 
@@ -33,6 +33,8 @@ def _coerce_registry_result(
     default_outcome: str,
     raw: NodeReturn[BaseModel] | BaseModel,
 ) -> dict[str, Any]:
+    if raw is None and output_model is Nothing:
+        return {"outcome": default_outcome, "output": {}}
     if isinstance(raw, NodeReturn):
         if not isinstance(raw.output, output_model):
             raise TypeError(
@@ -69,7 +71,12 @@ class NodeSpec(Generic[InputT, OutputT]):
         self,
         payload: InputT,
         ctx: RuntimeContext | None = None,
-    ) -> NodeReturn[OutputT] | OutputT | Awaitable[NodeReturn[OutputT] | OutputT]:
+    ) -> (
+        NodeReturn[OutputT]
+        | OutputT
+        | None
+        | Awaitable[NodeReturn[OutputT] | OutputT | None]
+    ):
         if self.accepts_context:
             if ctx is None:
                 raise TypeError(f"node {self.name!r} requires RuntimeContext")