What causes "No class found with the name {kind!r}, known classes are: {msg_str}" in docling-project/docling?

Import the plugin module that registers the model class before creating the instance.

docling-project/docling · error · RuntimeError

No class found with the name {kind!r}, known classes are: {m

Error message

No class found with the name {kind!r}, known classes are:
{msg_str}

What it means

Raised as RuntimeError by BaseFactory.create_instance when the concrete options type passed to the factory has no registered implementation class. The factory maps options classes to model classes; a KeyError on type(options) means no plugin providing that options class was ever registered in this process.

Source

Thrown at docling/models/factories/base_factory.py:59

            names={kind: kind for kind in self.registered_kind},
            type=str,
            module=__name__,
        )

    @property
    def classes(self):
        return self._classes

    @property
    def registered_meta(self):
        return self._meta

    def create_instance(self, options: BaseOptions, **kwargs) -> A:
        try:
            _cls = self._classes[type(options)]
            return _cls(options=options, **kwargs)
        except KeyError:
            raise RuntimeError(self._err_msg_on_class_not_found(options.kind))

    def create_options(self, kind: str, *args, **kwargs) -> BaseOptions:
        for opt_cls, _ in self._classes.items():
            if opt_cls.kind == kind:
                return opt_cls(*args, **kwargs)
        raise RuntimeError(self._err_msg_on_class_not_found(kind))

    def _err_msg_on_class_not_found(self, kind: str):
        msg = []

        for opt, cls in self._classes.items():
            msg.append(f"\t{opt.kind!r} => {cls!r}")

        msg_str = "\n".join(msg)

        return f"No class found with the name {kind!r}, known classes are:\n{msg_str}"

    def register(self, cls: Type[A], plugin_name: str, plugin_module_name: str):

View on GitHub (pinned to 61d76f1ff3)

Solutions

Import the plugin module that registers the model class before creating the instance.
Install the plugin package (e.g. uv add / pip install the plugin) so its entry point or import side effect runs.
Check factory.registered_meta / classes to confirm which kinds are actually available, and enable plugin loading (load_from_plugins) if the plugin ships via entry points.

Example fix

# before
options = MyModelOptions(...)  # from a plugin
model = factory.create_instance(options)  # RuntimeError

# after
import my_docling_plugin  # registers the class on import
model = factory.create_instance(options)

Defensive patterns

Strategy: validation

Validate before calling

factory = MyFactory.getInstance()
if type(options) not in factory.classes:
    raise SystemExit(
        f'{type(options).__name__} not registered; known kinds: '
        + ', '.join(opt.kind for opt in factory.classes)
    )

Type guard

def is_registered(factory, options) -> bool:
    return type(options) in factory.classes

Try / catch

try:
    model = factory.create_instance(options)
except RuntimeError as e:
    if 'No class found' in str(e):
        import my_docling_plugin  # noqa: F401 ensure registration
        model = factory.create_instance(options)
    else:
        raise

Prevention

Import plugin modules once at application startup, before any factory use.
Print factory.registered_meta during onboarding to confirm plugin discovery.
Pin plugin versions so registration side effects cannot silently disappear.

When it happens

Trigger: Calling create_instance(options) where type(options) was never registered via factory.register() or discovered by load_from_plugins(). Typical when the plugin package that defines the model implementation was never imported or installed.

Common situations: Using a third-party Docling model plugin without importing its module (registration usually happens at import time); plugin discovery disabled or restricted (allow_external_plugins=False); plugin package missing from the environment.

Related errors

AI-assisted analysis of docling-project/docling@61d76f1ff3 (2026-08-14). Data as JSON: /api/errors/51dbbb1bde12d59b. Report an issue: GitHub.