vipr.plugins.discovery package

Submodules

vipr.plugins.discovery.controllers module

Discovery Controller

Provides commands for interacting with the Discovery system.

class vipr.plugins.discovery.controllers.DiscoveryController(*args: Any, **kw: Any)

Bases: ArgparseController

Controller for Discovery functions.

class Meta

Bases: object

description = 'Manages the VIPR Component Discovery system'

help = 'Tools for managing VIPR components'

label = 'discovery'

stacked_on = 'base'

stacked_type = 'nested'

components()

Lists all available components. No input required for this GET endpoint.

data_loaders()

Lists all available data loaders.

filter_types()

Lists all available filter types.

filters()

Lists all available filters.

hook_types()

Lists all available hook types.

hooks()

Lists all available hooks.

model_loaders()

Lists all available model loaders.

plugins()

Lists all available plugins by type.

postprocessors()

Lists all available postprocessing handlers.

predictors()

Lists all available prediction handlers.

vipr.plugins.discovery.decorators module

Decorator-based discovery for components.

This module provides decorators for registering hooks, filters, data loaders, and other components with the central discovery system.

vipr.plugins.discovery.decorators.clear_discovery()

Clear all registered components (mainly for testing).

vipr.plugins.discovery.decorators.discover_component(registry: dict[str, list[dict[str, Any]]], component_type: str, registry_name: str, weight: int = 0, enabled_in_config: bool = True, **parameters)

Generic decorator to register a component (hook or filter).

Parameters:

• registry – The registry to add to (hooks_registry or filters_registry)

• component_type – The type of component (e.g. ‘POST_ARGUMENT_PARSING’)

• registry_name – Name for logging (“hook” or “filter”)

• weight – The execution priority (lower values run first)

• enabled_in_config – Whether the component is enabled in generated configs by default

• **parameters – Additional parameters for the component

Returns:

The decorator function

vipr.plugins.discovery.decorators.discover_data_loader(handler_name: str, parameters: type[BaseModel] | None = None)

Decorator to register a data loader.

Parameters:

• handler_name – The name/label of the data loader

• parameters – Pydantic model class describing handler parameters

Returns:

The decorated class

vipr.plugins.discovery.decorators.discover_filter(filter_type: str, weight: int = 0, enabled_in_config: bool = True, **parameters)

Decorator to register a filter.

Parameters:

• filter_type – The type of filter (e.g. ‘INFERENCE_PREPROCESS_PRE_FILTER’)

• weight – The execution priority of the filter (lower values run first)

• enabled_in_config – Whether the filter is enabled in generated configs by default

• **parameters – Additional parameters for the filter. parameters must be type[BaseModel] when provided.

Returns:

The decorated function

vipr.plugins.discovery.decorators.discover_hook(hook_type: str, weight: int = 0, enabled_in_config: bool = True, **parameters)

Decorator to discover a hook.

Parameters:

• hook_type – The type of hook (e.g. ‘POST_ARGUMENT_PARSING’)

• weight – The execution priority of the hook (lower values run first)

• enabled_in_config – Whether the hook is enabled in generated configs by default

• **parameters – Additional parameters for the hook. parameters must be type[BaseModel] when provided.

Returns:

The decorated function

vipr.plugins.discovery.decorators.discover_model_loader(handler_name: str, parameters: type[BaseModel] | None = None)

Decorator to register a model loader.

Parameters:

• handler_name – The name/label of the model loader

• parameters – Pydantic model class describing handler parameters

Returns:

The decorated class

vipr.plugins.discovery.decorators.discover_postprocessor(handler_name: str, parameters: type[BaseModel] | None = None)

Decorator to register a postprocessor.

Parameters:

• handler_name – The name/label of the postprocessor

• parameters – Pydantic model class describing handler parameters

Returns:

The decorated class

vipr.plugins.discovery.decorators.discover_predictor(handler_name: str, parameters: type[BaseModel] | None = None)

Decorator to register a predictor.

Parameters:

• handler_name – The name/label of the predictor

• parameters – Pydantic model class describing handler parameters

Returns:

The decorated class

vipr.plugins.discovery.decorators.get_discovered_components(registry: dict[str, list[dict[str, Any]]], component_type: str | None = None, plugin_name: str | list[str] | None = None) → dict[str, list[dict[str, Any]]]

Get registered components, filtered by type and/or plugin(s).

Parameters:

• registry – The registry to search in

• component_type – Optional component type to filter by

• plugin_name – Optional plugin name(s) to filter by. Can be a string or list of strings.

vipr.plugins.discovery.decorators.get_discovered_data_loaders(plugin_name: str | list[str] | None = None) → list[dict[str, Any]]

Get registered data loaders, optionally filtered by plugin(s).

Parameters:

plugin_name – Optional plugin name or list of plugin names to filter by

Returns:

list of data loader definitions

vipr.plugins.discovery.decorators.get_discovered_filters(filter_type: str | None = None, plugin_name: str | list[str] | None = None) → dict[str, list[dict[str, Any]]]

Get registered filters, optionally filtered by type and/or plugin(s).

Parameters:

• filter_type – Optional filter type to filter by

• plugin_name – Optional plugin name or list of plugin names to filter by

vipr.plugins.discovery.decorators.get_discovered_hooks(hook_type: str | None = None, plugin_name: str | list[str] | None = None) → dict[str, list[dict[str, Any]]]

Get registered hooks, optionally filtered by type and/or plugin(s).

Parameters:

• hook_type – Optional hook type to filter by

• plugin_name – Optional plugin name or list of plugin names to filter by

vipr.plugins.discovery.decorators.get_discovered_model_loaders(plugin_name: str | list[str] | None = None) → list[dict[str, Any]]

Get registered model loaders, optionally filtered by plugin(s).

Parameters:

plugin_name – Optional plugin name or list of plugin names to filter by

Returns:

list of model loader definitions

vipr.plugins.discovery.decorators.get_discovered_postprocessors(plugin_name: str | list[str] | None = None) → list[dict[str, Any]]

Get registered postprocessors, optionally filtered by plugin(s).

Parameters:

plugin_name – Optional plugin name or list of plugin names to filter by

Returns:

list of postprocessor definitions

vipr.plugins.discovery.decorators.get_discovered_predictors(plugin_name: str | list[str] | None = None) → list[dict[str, Any]]

Get registered predictors, optionally filtered by plugin(s).

Parameters:

plugin_name – Optional plugin name or list of plugin names to filter by

Returns:

list of predictor definitions

vipr.plugins.discovery.models module

Response models for the Registry API endpoints.

This module provides Pydantic models representing the various response formats used by the registry controller endpoints.

class vipr.plugins.discovery.models.CallbackEntry(*, enabled: bool, weight: int, method: str, class_: str, parameters: Dict[str, Any] | None = None, parameters_schema: Dict[str, Any] | None = None)

Bases: BaseModel

Base model for hook and filter entries.

class_: str

enabled: bool

classmethod map_enabled_field(data: Any) → Any

Accept both ‘enabled’ and ‘enabled_in_config’ fields.

This validator maps the internal registry field ‘enabled_in_config’ to the API/YAML field ‘enabled’ for backwards compatibility.

method: str

model_config: ClassVar[ConfigDict] = {}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

parameters: Dict[str, Any] | None

parameters_schema: Dict[str, Any] | None

weight: int

class vipr.plugins.discovery.models.ComponentsResponse(*, model_loaders: List[HandlerEntry], data_loaders: List[HandlerEntry], hooks: Dict[str, List[CallbackEntry]], filters: Dict[str, List[CallbackEntry]], predictors: List[HandlerEntry])

Bases: BaseModel

Response model for the components endpoint.

data_loaders: List[HandlerEntry]

filters: Dict[str, List[CallbackEntry]]

hooks: Dict[str, List[CallbackEntry]]

model_config: ClassVar[ConfigDict] = {}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

model_loaders: List[HandlerEntry]

predictors: List[HandlerEntry]

class vipr.plugins.discovery.models.ComponentsSummary(*, model_loaders: int, data_loaders: int, hooks: int, filters: int, predictors: int)

Bases: BaseModel

Summary of component counts by type.

data_loaders: int

filters: int

hooks: int

model_config: ClassVar[ConfigDict] = {}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

model_loaders: int

predictors: int

class vipr.plugins.discovery.models.HandlerEntry(*, handler: str, class_: str, parameters: ~typing.Dict[str, ~typing.Any] = <factory>, parameters_schema: ~typing.Dict[str, ~typing.Any] | None = None)

Bases: BaseModel

Base model for model loaders, data loaders, and predictors.

class_: str

handler: str

model_config: ClassVar[ConfigDict] = {}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

parameters: Dict[str, Any]

parameters_schema: Dict[str, Any] | None

class vipr.plugins.discovery.models.PluginSummaryResponse(*, components_summary: ComponentsSummary)

Bases: BaseModel

Response model for plugin summary endpoint.

components_summary: ComponentsSummary

model_config: ClassVar[ConfigDict] = {}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

Module contents

VIPR Registry Plugin

This plugin provides a registry system for VIPR components: - Model Loaders - Data Loaders - Hooks - Filters

It enables listing available components and generating configuration files.

vipr.plugins.discovery.load(app)

Loads the Registry plugin