Source code for molten.dependency_injection

# This file is a part of molten.
#
# Copyright (C) 2018 CLEARTYPE SRL <[email protected]>
#
# molten is free software; you can redistribute it and/or modify it
# under the terms of the GNU Lesser General Public License as published by
# the Free Software Foundation, either version 3 of the License, or (at
# your option) any later version.
#
# molten is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
# FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
# License for more details.
#
# You should have received a copy of the GNU Lesser General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.

import functools
from inspect import Parameter, signature
from typing import Any, Callable, Dict, Iterable, List, Optional, TypeVar, no_type_check

from typing_extensions import Protocol

from .errors import DIError

_T = TypeVar("_T", covariant=True)


[docs]class Component(Protocol[_T]): # pragma: no cover """The component protocol. Examples: >>> class DBComponent: ... is_cacheable = True ... is_singleton = True ... ... def can_handle_parameter(self, parameter: Parameter) -> bool: ... return parameter.annotation is DB ... ... def resolve(self, settings: Settings) -> DB: ... return DB(settings["database_dsn"]) """ @property def is_cacheable(self) -> bool: """If True, then the component will be cached within a resolver meaning that instances of the resolved component will be reused within a single request-response cycle. This should be True for most components. Defaults to True. """ @property def is_singleton(self) -> bool: """If True, then the component will be treated as a singleton and cached forever after its first use. Defaults to False. """
[docs] def can_handle_parameter(self, parameter: Parameter) -> bool: """Returns True when parameter represents the desired component. """
[docs] @no_type_check def resolve(self) -> _T: """Returns an instance of the component. """
[docs]class DependencyInjector: """The dependency injector maintains component state and instantiates the resolver. Parameters: components: The list of components that are used to resolve functions' dependencies. """ __slots__ = [ "components", "singletons", ] components: List[Component[Any]] singletons: Dict[Component[Any], Any] def __init__(self, components: List[Component[Any]], singletons: Optional[Dict[Component[Any], Any]] = None) -> None: self.components = components or [] self.singletons = singletons or {} def is_singleton(component: Component[Any]) -> bool: return getattr(component, "is_singleton", False) and component not in self.singletons for component in components: if is_singleton(component): resolver = self.get_resolver() resolved_component = resolver.resolve(component.resolve) self.singletons[component] = resolved_component() # Also register any singleton components that were # instantiated during the resolving process of the # current component. This is necessary so that # singletons don't get instantiated by each dependent # component in part. for component, instance in resolver.instances.items(): if is_singleton(component): self.singletons[component] = instance
[docs] def get_resolver(self, instances: Optional[Dict[Any, Any]] = None) -> "DependencyResolver": """Get the resolver for this Injector. """ return DependencyResolver( self.components, {**self.singletons, **(instances or {})}, )
[docs]class DependencyResolver: """The resolver does the work of actually filling in all of a function's dependencies. """ __slots__ = [ "components", "instances", ] def __init__(self, components: List[Component[Any]], instances: Dict[Component[Any], Any]) -> None: self.components = components[:] self.instances = instances
[docs] def add_component(self, component: Component[Any]) -> None: """Add a component to this resolver without adding it to the base dependency injector. This is useful for runtime-built components like RouteParamsComponent. """ self.components.append(component)
[docs] def resolve( self, fn: Callable[..., Any], resolving_parameter: Optional[Parameter] = None, ) -> Callable[..., Any]: """Resolve a function's dependencies. """ @functools.wraps(fn) def resolved_fn(**params: Any) -> Any: for parameter in _get_parameters(fn): if parameter.name in params: continue # When Parameter is requested then we assume that the # caller actually wants the parameter that resolved the # current component that's being resolved. See QueryParam # or Header components for an example. if parameter.annotation is Parameter: params[parameter.name] = resolving_parameter continue # When a DependencyResolver is requested, then we assume # the caller wants this instance. if parameter.annotation is DependencyResolver: params[parameter.name] = self continue # If our instances contains an exact match for a type, # then we return that. This is used to inject the current # Request object among other things. try: params[parameter.name] = self.instances[parameter.annotation] continue except KeyError: pass for component in self.components: if component.can_handle_parameter(parameter): try: params[parameter.name] = self.instances[component] except KeyError: factory = self.resolve(component.resolve, resolving_parameter=parameter) params[parameter.name] = instance = factory() if getattr(component, "is_cacheable", True): self.instances[component] = instance break else: raise DIError(f"cannot resolve parameter {parameter} of function {fn}") return fn(**params) return resolved_fn
@functools.lru_cache(maxsize=128) def _get_parameters(fn: Callable[..., Any]) -> Iterable[Parameter]: # A significant amount of time is spent getting handlers' params. # Since they never change, it should be safe to just cache 'em. return signature(fn).parameters.values()