`ToolkitRegistry`

class openff.toolkit.utils.toolkits.ToolkitRegistry(toolkit_precedence: list[type[ToolkitWrapper]] | None = None, exception_if_unavailable: bool = True, _register_imported_toolkit_wrappers: bool = False)[source]

A registry and precedence list for ToolkitWrapper objects.

ToolkitRegistry allows the OpenFF Toolkit to provide a concise, universal API that can call out to other well-established libraries rather than re-implement algorithms with well-established implementations, while still giving users control over dependencies. It contains a list of ToolkitWrapper objects, each of which provides some collection of methods that may be requested from the registry. The call() method takes a name and calls the method of that name on each wrapper until it finds a working implementation, whose result it returns. For details on how this search is conducted and what counts as a working implementation, see that method’s API docs.

Examples

>>> from openff.toolkit.utils.toolkits import ToolkitRegistry
>>> toolkit_precedence = [OpenEyeToolkitWrapper, RDKitToolkitWrapper, AmberToolsToolkitWrapper]
>>> toolkit_registry = ToolkitRegistry(toolkit_precedence)
>>> toolkit_registry
<ToolkitRegistry containing OpenEye Toolkit, The RDKit, AmberTools>

>>> toolkits = [OpenEyeToolkitWrapper, RDKitToolkitWrapper, AmberToolsToolkitWrapper, BuiltInToolkitWrapper]
>>> toolkit_registry = ToolkitRegistry(toolkit_precedence=toolkits)
>>> toolkit_registry
<ToolkitRegistry containing OpenEye Toolkit, The RDKit, AmberTools, Built-in Toolkit>

Retrieve the global singleton toolkit registry, which is created when this module is imported from all available toolkits:

>>> from openff.toolkit.utils.toolkits import GLOBAL_TOOLKIT_REGISTRY
>>> GLOBAL_TOOLKIT_REGISTRY
<ToolkitRegistry containing OpenEye Toolkit, The RDKit, AmberTools, Built-in Toolkit>

Note that this will contain different ToolkitWrapper objects based on what toolkits are currently installed.

Call a method on the registered toolkit wrapper with the highest precedence:

>>> molecule = toolkit_registry.call('from_smiles', 'Cc1ccccc1')

For more, see the call() method.

Warning

This API is experimental and subject to change.

__init__(toolkit_precedence: list[type[ToolkitWrapper]] | None = None, exception_if_unavailable: bool = True, _register_imported_toolkit_wrappers: bool = False)[source]

Create an empty toolkit registry.

Parameters:

toolkit_precedence – List of toolkit wrapper classes, in order of desired precedence when performing molecule operations. If None, no toolkits will be registered.
exception_if_unavailable – If True, an exception will be raised if the toolkit is unavailable
_register_imported_toolkit_wrappers – If True, will attempt to register all imported ToolkitWrapper subclasses that can be found in the order of toolkit_precedence, if specified. If toolkit_precedence is not specified, the default order is [OpenEyeToolkitWrapper, RDKitToolkitWrapper, AmberToolsToolkitWrapper, BuiltInToolkitWrapper].

Methods

`__init__`([toolkit_precedence, ...])	Create an empty toolkit registry.
`add_toolkit`(toolkit_wrapper)	Append a ToolkitWrapper onto the list of toolkits in this ToolkitRegistry
`call`(method_name, *args[, raise_exception_types])	Execute a method with the first registered toolkits that supports it.
`deregister_toolkit`(toolkit_wrapper)	Remove a ToolkitWrapper from the list of toolkits in this ToolkitRegistry
`register_toolkit`(toolkit_wrapper[, ...])	Register the provided toolkit wrapper class, instantiating an object of it.
`resolve`(method_name)	Get a method with the given name from the first registered toolkit that provides it.

Attributes

`registered_toolkit_versions`	Return a dict containing the version of each registered toolkit.
`registered_toolkits`	List registered toolkits.

property registered_toolkits: list[ToolkitWrapper]

List registered toolkits.

Warning

This API is experimental and subject to change.

Returns:: toolkits

property registered_toolkit_versions: dict[str, str]

Return a dict containing the version of each registered toolkit.

Warning

This API is experimental and subject to change.

Returns:: toolkit_versions – A dictionary mapping names and versions of wrapped toolkits

call(method_name: str, *args, raise_exception_types: list[type[Exception]] | None = None, **kwargs)[source]

Execute a method with the first registered toolkits that supports it.

This method searches the registry’s precedence list for the first wrapper that has an attribute named method_name and attempts to call it as a method using the arguments in *args and **kwargs. If that method raises no exception, its return value is returned.

By default, if a wrapper with an appropriately-named method raises an exception of any type, then iteration over the registered toolkits stops early and that exception is raised. To limit this behavior to only certain exceptions and otherwise continue iteration, customize this behavior using the optional raise_exception_types keyword argument. If iteration finishes without finding a wrapper that can successfully call the requested method, a ValueError is raised, containing a message listing the registered toolkits and any exceptions that occurred.

Parameters:

method_name – The name of the method to execute
raise_exception_types – A list of exception-derived types to catch and raise immediately. If None, the first exception encountered will be raised. To ignore all exceptions, set this to the empty list [].

Raises:

ValueError – If no suitable toolkit wrapper was found in the registry.
Exception – Other forms of exceptions are possible if raise_exception_types is specified. These are defined by the ToolkitWrapper method being called.

Examples

Create a molecule, and call the toolkit to_smiles() method directly

>>> from openff.toolkit import (
...     Molecule,
...     ToolkitRegistry,
...     AmberToolsToolkitWrapper,
...     RDKitToolkitWrapper,
... )
>>> molecule = Molecule.from_smiles('Cc1ccccc1')
>>> toolkit_registry = ToolkitRegistry([
...     AmberToolsToolkitWrapper,
...     RDKitToolkitWrapper,
... ])
>>> smiles = toolkit_registry.call('to_smiles', molecule)

Stop if a partial charge assignment method encounters an error during the partial charge calculation (ChargeCalculationError), but proceed to the next wrapper for any other exception such as the wrapper not supporting the charge method:

>>> from openff.toolkit import GLOBAL_TOOLKIT_REGISTRY
>>> from openff.toolkit.utils.exceptions import ChargeCalculationError
>>> GLOBAL_TOOLKIT_REGISTRY.call(
...     "assign_partial_charges",
...     molecule=Molecule.from_smiles('C'),
...     partial_charge_method="gasteiger",
...     raise_exception_types=[ChargeCalculationError],
... )

Calling a method that exists on none of the wrappers raises a ValueError:

>>> from openff.toolkit import ToolkitRegistry, RDKitToolkitWrapper
>>> toolkit_registry = ToolkitRegistry([RDKitToolkitWrapper])
>>> toolkit_registry.call("there_is_no_spoon") 
Traceback (most recent call last):
  ...
ValueError: No registered toolkits can provide the capability
"there_is_no_spoon" for args "()" and kwargs "{}"
Available toolkits are: [ToolkitWrapper around The RDKit version ...]

register_toolkit(toolkit_wrapper: ToolkitWrapper | type[ToolkitWrapper], exception_if_unavailable: bool = True)[source]

Warning

This API is experimental and subject to change.

Parameters:

toolkit_wrapper – The toolkit wrapper to register or its class.
exception_if_unavailable – If True, an exception will be raised if the toolkit is unavailable

deregister_toolkit(toolkit_wrapper: ToolkitWrapper)[source]

Remove a ToolkitWrapper from the list of toolkits in this ToolkitRegistry

Warning

This API is experimental and subject to change.

Parameters:

toolkit_wrapper – The toolkit wrapper to remove from the registry

Raises:

InvalidToolkitError – If toolkit_wrapper is not a ToolkitWrapper or subclass
ToolkitUnavailableException – If toolkit_wrapper is not found in the registry

add_toolkit(toolkit_wrapper: ToolkitWrapper)[source]

Append a ToolkitWrapper onto the list of toolkits in this ToolkitRegistry

Warning

This API is experimental and subject to change.

Parameters:: toolkit_wrapper – The ToolkitWrapper object to add to the list of registered toolkits
Raises:: InvalidToolkitError – If toolkit_wrapper is not a ToolkitWrapper or subclass

resolve(method_name: str) → Callable[source]

Get a method with the given name from the first registered toolkit that provides it.

Resolve the requested method name by checking all registered toolkits in order of precedence for one that provides the requested method. Note that this may not be the method used by call() if the raise_exception_types argument is passed.

Parameters:: method_name – The name of the method to resolve
Returns:: method – The method of the first registered toolkit that provides the requested method name
Raises:: NotImplementedError – if the requested method cannot be found among the registered toolkits

Examples

Create a molecule, and call the toolkit to_smiles() method directly

>>> from openff.toolkit import Molecule
>>> molecule = Molecule.from_smiles('Cc1ccccc1')
>>> toolkit_registry = ToolkitRegistry([
...     OpenEyeToolkitWrapper,
...     RDKitToolkitWrapper,
...     AmberToolsToolkitWrapper
... ])
>>> method = toolkit_registry.resolve('to_smiles')
>>> smiles = method(molecule)