class-resolver

What it is and what it does

class-resolver provides a ClassResolver object that maintains a registry of classes and resolves them by string name, class reference, or pre-instantiated object. It accepts flexible input formats (lowercase strings, stylized names, class objects, or instances) and instantiates them with optional constructor arguments. The package is designed to eliminate boilerplate dictionary-based class lookup patterns and hard-coded conditional logic, making it especially useful for building extensible machine learning models and plugin systems where users need to specify components via configuration strings.

The core use case is replacing manual if-elif chains or static lookup dictionaries with a declarative registry that handles case-insensitive string matching, class validation against a base type, and optional default instantiation. It includes contrib modules for common frameworks, reducing repetitive setup code across projects that need flexible component selection.

Use it for:

• Build extensible neural network architectures where activation functions are selected by string configuration.
• Create plugin systems where users register custom classes and reference them by name in config files.
• Integrate with hyperparameter optimization libraries that require serializable string representations of class choices.
• Reduce boilerplate in factory patterns by replacing manual class-to-string mappings with automatic registry-based lookup.
• Support both programmatic (class/instance) and declarative (string) class selection in the same API.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Dynamically lookup and instantiate classes by string name, class object, or instance, with optional constructor arguments.

Yes. The package is actively maintained, has no known vulnerabilities, minimal dependencies, and solves a genuine pain point in extensible system design. It's particularly valuable if you're building configurable ML models, plugin architectures, or any system where users need to specify classes via strings. The low install friction and permissive license make adoption straightforward.

Install

class-resolver on PyPI

pip install class-resolver

uv add class-resolver

poetry add class-resolver

Installing class-resolver

Before you install

Low friction: pure Python wheel with only typing-extensions as a runtime dependency. Active maintenance with latest release on 2025-08-28.

License in practice

MIT license (permissive) means you can use, modify, and distribute this package freely in commercial or private projects with minimal restrictions.

Quickstart

from class_resolver import ClassResolver

class Base: pass
class A(Base): pass
class B(Base): pass

resolver = ClassResolver([A, B], base=Base)
assert A == resolver.lookup('A')
assert A(name='hi') == resolver.make('A', {'name': 'hi'})

Requires Python 3.10 or later

Verify before relying

• Whether the package handles circular imports or late-binding class registration patterns
• Performance characteristics when resolving from large class registries
• Whether contrib.torch module requires PyTorch to be installed or is optional

Package facts

Evidence: class_resolver-0.7.1-py3-none-any.whl

Keywords: snekpack, cookiecutter, development tool, configurability

Tags

More Application Frameworks packages