`utils.llm_pii_utils`#

Module Contents#

Classes#

EntitySpan

Functions#

`find_entity_spans`	Find the start and end indexes for each entity in the given text.
`fix_overlaps`	Handle overlaps in entity spans
`get_system_prompt`
`redact`	Redact given entities from the original text
`validate_entity`	Validate entity
`validate_keys`	Validate that keys in entity dict match schema

Data#

`JSON_SCHEMA`
`PII_LABELS`

API#

class utils.llm_pii_utils.EntitySpan#

end_position: int#: None

entity_type: str#: None

start_position: int#: None

utils.llm_pii_utils.JSON_SCHEMA#: None

utils.llm_pii_utils.PII_LABELS#: [‘medical_record_number’, ‘location’, ‘address’, ‘ssn’, ‘date_of_birth’, ‘date_time’, ‘name’, ‘email…

utils.llm_pii_utils.find_entity_spans( text: str, entities: list[dict[str, str]], ) → list[utils.llm_pii_utils.EntitySpan]#

Find the start and end indexes for each entity in the given text.

Args: text (str): The input text string. entities (list): A list of entities, where each entity is a dictionary containing the entity text and its type.

Returns: list: A list of EntitySpan objects, where each contains the entity type, start position, and end position.

utils.llm_pii_utils.fix_overlaps( spans: list[utils.llm_pii_utils.EntitySpan], ) → list[utils.llm_pii_utils.EntitySpan]#: Handle overlaps in entity spans

utils.llm_pii_utils.get_system_prompt(pii_labels: list[str] = PII_LABELS) → str#

utils.llm_pii_utils.redact(full_text: str, pii_entities: list[dict[str, str]]) → str#: Redact given entities from the original text

utils.llm_pii_utils.validate_entity( entity: dict[str, str], text: str, min_length: int = 2, ) → bool#: Validate entity

utils.llm_pii_utils.validate_keys(entity_dict: dict) → bool#: Validate that keys in entity dict match schema

utils.llm_pii_utils#