nemoguardrails.manifests.config_schema

Building blocks for declaring typed, per-rail configuration.

Rails define their configuration as a RailConfigBaseModel (a Pydantic base that ignores unknown keys) with fields built via rail_field. RailConfigSpec describes a single configuration field: its annotation, the Pydantic FieldInfo, and any named values it exports to the runtime.

Module Contents

Classes

Name	Description
`RailConfigBaseModel`	Base model for typed per-rail configuration sections.
`RailConfigSpec`	Typed field definition and exported values for a rail configuration.

Functions

Name	Description
`rail_field`	Create Pydantic field metadata for a rail configuration field.

Data

__all__

API

class nemoguardrails.manifests.config_schema.RailConfigBaseModel()

Bases: _PydanticBaseModel

Base model for typed per-rail configuration sections.

model_config

= _ConfigDict(extra='ignore')

class nemoguardrails.manifests.config_schema.RailConfigSpec(
    annotation: typing.Any,
    field_info: pydantic.fields.FieldInfo,
    exports: typing.Dict[str, typing.Any] = dict(),
    key: typing.Optional[str] = None
)

Dataclass

Typed field definition and exported values for a rail configuration.

exports

Dict[str, Any] = field(default_factory=dict)

field_info

FieldInfo

key

Optional[str] = field(default=None, kw_only=True)

nemoguardrails.manifests.config_schema.rail_field(
    args: typing.Any = (),
    kwargs: typing.Any = {}
) -> pydantic.fields.FieldInfo

Create Pydantic field metadata for a rail configuration field.

nemoguardrails.manifests.config_schema.__all__ = ['Discriminator', 'Field', 'PrivateAttr', 'RailConfigBaseModel', 'RailConfigSpec...