guardrails_ai.types¶

Submodules¶

Classes¶

`CreateGuardRequest`	The required request body to created a Guard in the Guardrails API.
`ErrorSpan`	Character-level span within validated text that caused a validation failure.
`FailResult`	The output of a validator when validation fails.
`Guard`	A configured validation pipeline retrieved from the Guardrails API.
`JSONSchema`	A strongly typed representation of JSON Schema Draft 2020-12.
`OnFail`	OnFail is an Enum that represents the different actions that can
`Outcome`	str(object='') -> str
`PassResult`	PassResult is the output type of Validator.validate when validation
`ReAsk`	Represents a pending reask when validation fails and retries are exhausted.
`ValidationOutcome`	The output from a Guard execution.
`ValidationResult`	ValidationResult is the output type of Validator.validate and the
`ValidationSummary`	Per-validator result produced during a Guard execution.
`Validator`	A validator attached to a Guard, including its configuration.

Package Contents¶

class guardrails_ai.types.CreateGuardRequest¶

Bases: pydantic.BaseModel

The required request body to created a Guard in the Guardrails API.

description: str | None¶

model_config¶

name: str¶

output_schema: guardrails_ai.types.json_schema_2020_12.JSONSchema¶

validators: List[guardrails_ai.types.validator.Validator]¶

class guardrails_ai.types.ErrorSpan¶

Bases: pydantic.BaseModel

Character-level span within validated text that caused a validation failure.

Useful for pinpointing failures when validating large chunks of text or streaming output with varying chunk sizes.

end: int¶

model_config¶

reason: str¶

start: int¶

class guardrails_ai.types.FailResult¶

Bases: guardrails_ai.types.validation_result.ValidationResult

The output of a validator when validation fails.

classmethod deserialize_outcome(outcome: str | None) → Literal[guardrails_ai.types.validation_result.Outcome.FAIL]¶

error_message: str¶

error_spans: List[guardrails_ai.types.error_span.ErrorSpan] | None¶

fix_value: Any | None¶

outcome: guardrails_ai.types.validation_result.Outcome¶

class guardrails_ai.types.Guard¶

Bases: CreateGuardRequest

A configured validation pipeline retrieved from the Guardrails API.

id: str¶

model_config¶

class guardrails_ai.types.JSONSchema¶

Bases: pydantic.BaseModel

A strongly typed representation of JSON Schema Draft 2020-12.

JSON Schema can be either a boolean or an object with various properties. When boolean: - true: validates any instance - false: validates no instance

When object: contains various keywords from different vocabularies.

validate_conditional() → JSONSchema¶: Validate that then/else are only used with if.

validate_contains_constraints() → JSONSchema¶: Validate that minContains and maxContains are used with contains.

validate_length_constraints() → JSONSchema¶: Validate min/max length constraints.

validate_numeric_constraints() → JSONSchema¶: Validate numeric constraints.

classmethod validate_type(v: StringOrStringArray | None) → StringOrStringArray | None¶: Validate that type values are one of the allowed JSON Schema types.

additional_properties: SchemaValue | None¶

all_of: List[SchemaValue] | None¶

anchor: str | None¶

any_of: List[SchemaValue] | None¶

comment: str | None¶

const: Any | None¶

contains: SchemaValue | None¶

content_encoding: str | None¶

content_media_type: str | None¶

content_schema: SchemaValue | None¶

default: Any | None¶

definitions: Dict[str, SchemaValue] | None¶

defs: Dict[str, SchemaValue] | None¶

dependencies: Dict[str, SchemaValue | List[str]] | None¶

dependent_required: Dict[str, List[str]] | None¶

dependent_schemas: Dict[str, SchemaValue] | None¶

deprecated: bool | None¶

description: str | None¶

dynamic_anchor: str | None¶

dynamic_ref: str | None¶

else_: SchemaValue | None¶

enum: List[Any] | None¶

examples: List[Any] | None¶

exclusive_maximum: float | int | None¶

exclusive_minimum: float | int | None¶

format: str | None¶

id: str | None¶

if_: SchemaValue | None¶

items: SchemaValue | None¶

max_contains: int | None¶

max_items: int | None¶

max_length: int | None¶

max_properties: int | None¶

maximum: float | int | None¶

min_contains: int | None¶

min_items: int | None¶

min_length: int | None¶

min_properties: int | None¶

minimum: float | int | None¶

model_config¶

multiple_of: float | int | None¶

not_: SchemaValue | None¶

one_of: List[SchemaValue] | None¶

pattern: str | None¶

pattern_properties: Dict[str, SchemaValue] | None¶

prefix_items: List[SchemaValue] | None¶

properties: Dict[str, SchemaValue] | None¶

property_names: SchemaValue | None¶

read_only: bool | None¶

recursive_anchor: bool | None¶

recursive_ref: str | None¶

ref: str | None¶

required: List[str] | None¶

schema_: str | None¶

then: SchemaValue | None¶

title: str | None¶

type: StringOrStringArray | None¶

unevaluated_items: SchemaValue | None¶

unevaluated_properties: SchemaValue | None¶

unique_items: bool | None¶

vocabulary: Dict[str, bool] | None¶

write_only: bool | None¶

class guardrails_ai.types.OnFail¶

Bases: str, enum.Enum

OnFail is an Enum that represents the different actions that can be taken when a validation fails.

REASK¶

On failure, Reask the LLM.

Type:: Literal[“reask”]

FIX¶

On failure, apply a static fix.

Type:: Literal[“fix”]

FILTER¶

On failure, filter out the invalid values.

Type:: Literal[“filter”]

REFRAIN¶

On failure, refrain from responding; return an empty value.

Type:: Literal[“refrain”]

NOOP¶

On failure, do nothing.

Type:: Literal[“noop”]

EXCEPTION¶

On failure, raise a ValidationError.

Type:: Literal[“exception”]

FIX_REASK¶

On failure, apply a static fix, check if the fixed value passed validation, if not then reask the LLM.

Type:: Literal[“fix_reask”]

CUSTOM¶

On failure, call a custom function with the invalid value and the FailResult’s from any validators run on the value.

Type:: Literal[“custom”]

Initialize self. See help(type(self)) for accurate signature.

CUSTOM = 'custom'¶

EXCEPTION = 'exception'¶

FILTER = 'filter'¶

FIX = 'fix'¶

FIX_REASK = 'fix_reask'¶

NOOP = 'noop'¶

REASK = 'reask'¶

REFRAIN = 'refrain'¶

class guardrails_ai.types.Outcome¶

Bases: str, enum.Enum

str(object=’’) -> str str(bytes_or_buffer[, encoding[, errors]]) -> str

Create a new string object from the given object. If encoding or errors is specified, then the object must expose a data buffer that will be decoded using the given encoding and error handler. Otherwise, returns the result of object.__str__() (if defined) or repr(object). encoding defaults to ‘utf-8’. errors defaults to ‘strict’.

Initialize self. See help(type(self)) for accurate signature.