robolab_motionplanning / robolab /core /utils /function_loader.py
yqi19's picture
Upload RoboLab motion-planning code only
81c7a5f verified
Raw
History Blame Contribute Delete
6.12 kB
# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
import importlib
import pkgutil
from functools import partial
from inspect import signature
from typing import Any, Callable
import robolab.core.utils.file_utils as file_utils
import robolab.core.utils.params_utils as params_utils
def get_callable_info(func: Callable) -> tuple[str, dict[str, Any]]:
"""
Extract function name from a callable function, and return the function name and pre-filled arguments.
"""
if isinstance(func, partial):
func_name = func.func.__name__
params = func.keywords
else:
func_name = func.__name__
params = func.__dict__
return func_name, params
def func_as_str(func: Callable) -> str:
"""
Convert a func function to a string.
"""
func_name, params = get_callable_info(func)
param_str = ', '.join(f"{k}={v}" for k, v in params.items())
return f"{func_name}({param_str})"
def load_callable_from_module(module_name: str | Any, name: str) -> Callable:
"""
Load a callable (function or class) from a given module.
Args:
module_name (str): The module to import.
name (str): The name of the callable object (function or class).
Returns:
Callable: The loaded callable object.
Raises:
ImportError: If the module or object cannot be found.
TypeError: If the object is not callable.
"""
try:
if isinstance(module_name, str):
module = importlib.import_module(module_name)
else:
module = module_name
obj = getattr(module, name)
if not callable(obj):
raise TypeError(f"'{name}' in module '{module_name}' is not callable.")
return obj
except (ModuleNotFoundError, AttributeError) as e:
raise ImportError(f"Cannot load '{name}' from module '{module_name}': {e}")
def search_function_in_module(
module, function_name: str, filename: str = None, function_params: dict = None
) -> Callable:
"""
The function that searches through the module (folder) and looks for the appropriate function.
If the function is not found in the appropriate folder, it raises a ValueError.
Example usage:
import module
callable_function = search_function_in_module(module, "function_name", myclass.py)
output = callable_function(**params)
Args:
module (imported module): _description_
function_name (str): name of the function.
filename (str, optional): The python file that contains the function. If none is provided, it will search through the entire folder to find the function with a matching name.Defaults to None.
function_params (dict, optional): optional params to the function. if provided, it will return a partial callable function.
Raises:
ValueError: _description_
Returns:
Callable: the callable policy function.
"""
func = None
if filename is not None:
if filename.endswith(".py"):
filename = file_utils.get_filename_without_extension(filename)
func = load_callable_from_module(
module.__name__ + "." + filename, function_name
)
else:
for importer, modname, ispkg in pkgutil.iter_modules(module.__path__):
func = load_callable_from_module(
module.__name__ + "." + modname, function_name
)
if func is not None:
break
if func is None:
raise ValueError(
f"No suitable function [function_name: '{function_name}', filename: '{filename}', module: {module.__name__}] found!"
)
if function_params is not None and isinstance(function_params, dict):
func = partial(func, **function_params)
return func
def load_callable_from_dict(
config: dict, prefill=False
) -> tuple[Callable, list, dict, str]:
"""
Dynamically load either a function or a class from a module based on the given configuration.
Args:
config (dict): A dictionary with the following structure:
{
"module": "module_name",
"function": "function_name",
"class": "ClassName",
"args": [arg1, arg2], # Optional for classes
"kwargs": {"param1": value1, "param2": value2} # Optional for both
}
prefill (bool): if True, prefills the class initialization or function with args and kwargs.
Returns:
tuple[Callable, list, dict, str]:
If type is "function", returns the callable function along with any args, kwargs, type
If type is "class", returns an instantiated class object along with any args, kwargs, type
"""
module_name = config.get("module")
args = config.get("args", [])
kwargs = config.get("kwargs", {})
params_utils.check_required_params_available(config, ["module"])
callable_type = params_utils.check_one_of_required_params_available(
config, ["function", "class"]
)
callable_name = config.get(callable_type)
fcn = load_callable_from_module(module_name, callable_name)
if prefill:
fcn = prefill_callable(fcn, args, kwargs)
return fcn, args, kwargs, callable_type
def prefill_callable(fcn: Callable | Any, args: list = None, kwargs: dict = None):
if kwargs or args:
if args is None:
args = ()
if kwargs is None:
kwargs = {}
fcn = partial(fcn, *args, **kwargs)
return fcn
def verify_callable_args_supplied(func: Callable, params: dict) -> tuple[bool, str]:
"""
Verify if all arguments are supplied to a callable function.
"""
p = partial(func, **params)
sig = signature(p.func)
try:
sig.bind_partial(*p.args, **(p.keywords or {}))
return True, None
except TypeError as e:
message = f"Error: {e}. Function {func.__name__} expects {list(sig.parameters.keys())} but got {p.args} and {p.keywords}."
return False, message