fix: make gpu-enabled simple function pickleable (#916)

fix: make gpu-enabled simple function picklable

Author: georgeh0

python/cocoindex/flow.py

@@ -105,18 +105,26 @@ def _spec_kind(spec: Any) -> str:
 
 def _transform_helper(
     flow_builder_state: _FlowBuilderState,
-    fn_spec: FunctionSpec,
+    fn_spec: FunctionSpec | Callable[..., Any],
     transform_args: list[tuple[Any, str | None]],
     name: str | None = None,
 ) -> DataSlice[Any]:
-    if not isinstance(fn_spec, FunctionSpec):
+    if isinstance(fn_spec, FunctionSpec):
+        kind = _spec_kind(fn_spec)
+        spec = fn_spec
+    elif callable(fn_spec) and (
+        op_kind := getattr(fn_spec, "__cocoindex_op_kind__", None)
+    ):
+        kind = op_kind
+        spec = op.EmptyFunctionSpec()
+    else:
         raise ValueError("transform() can only be called on a CocoIndex function")
 
     return _create_data_slice(
         flow_builder_state,
         lambda target_scope, name: flow_builder_state.engine_flow_builder.transform(
-            _spec_kind(fn_spec),
-            dump_engine_object(fn_spec),
+            kind,
+            dump_engine_object(spec),
             transform_args,
             target_scope,
             flow_builder_state.field_name_builder.build_name(
@@ -245,7 +253,7 @@ def for_each(
             f(scope)
 
     def transform(
-        self, fn_spec: op.FunctionSpec, *args: Any, **kwargs: Any
+        self, fn_spec: op.FunctionSpec | Callable[..., Any], *args: Any, **kwargs: Any
     ) -> DataSlice[Any]:
         """
         Apply a function to the data slice.
@@ -513,7 +521,7 @@ def add_source(
         )
 
     def transform(
-        self, fn_spec: FunctionSpec, *args: Any, **kwargs: Any
+        self, fn_spec: FunctionSpec | Callable[..., Any], *args: Any, **kwargs: Any
     ) -> DataSlice[Any]:
         """
         Apply a function to inputs, returning a DataSlice.

python/cocoindex/op.py

@@ -388,53 +388,39 @@ def _inner(cls: type[Executor]) -> type:
     return _inner
 
 
-class _EmptyFunctionSpec(FunctionSpec):
+class EmptyFunctionSpec(FunctionSpec):
     pass
 
 
 class _SimpleFunctionExecutor:
-    spec: Any
+    spec: Callable[..., Any]
 
     def prepare(self) -> None:
-        self.__call__ = self.spec.__call__
+        self.__call__ = staticmethod(self.spec)
 
 
-def function(**args: Any) -> Callable[[Callable[..., Any]], FunctionSpec]:
+def function(**args: Any) -> Callable[[Callable[..., Any]], Callable[..., Any]]:
     """
     Decorate a function to provide a function for an op.
     """
     op_args = OpArgs(**args)
 
-    def _inner(fn: Callable[..., Any]) -> FunctionSpec:
+    def _inner(fn: Callable[..., Any]) -> Callable[..., Any]:
         # Convert snake case to camel case.
-        op_name = "".join(word.capitalize() for word in fn.__name__.split("_"))
+        op_kind = "".join(word.capitalize() for word in fn.__name__.split("_"))
         sig = inspect.signature(fn)
-        full_name = f"{fn.__module__}.{fn.__qualname__}"
-
-        # An object that is both callable and can act as a FunctionSpec.
-        class _CallableSpec(_EmptyFunctionSpec):
-            __call__ = staticmethod(fn)
-
-            def __reduce__(self) -> str | tuple[Any, ...]:
-                return full_name
-
-        _CallableSpec.__name__ = op_name
-        _CallableSpec.__doc__ = fn.__doc__
-        _CallableSpec.__qualname__ = fn.__qualname__
-        _CallableSpec.__module__ = fn.__module__
-        callable_spec = _CallableSpec()
-
+        fn.__cocoindex_op_kind__ = op_kind  # type: ignore
         _register_op_factory(
             category=OpCategory.FUNCTION,
             expected_args=list(sig.parameters.items()),
             expected_return=sig.return_annotation,
             executor_factory=_SimpleFunctionExecutor,
-            spec_loader=lambda: callable_spec,
-            op_kind=op_name,
+            spec_loader=lambda: fn,
+            op_kind=op_kind,
             op_args=op_args,
         )
 
-        return callable_spec
+        return fn
 
     return _inner