def _add_cm_entity()

in src/engine/step1/_extraction_helpers.py [0:0]

• 30 lines of code
• 13 McCabe index (conditional complexity)

def _add_cm_entity(raw_entity, entities_by_category, seen_names):
    """Classifies and records names detected by CM based on CM category and type into tool's categories.


    Args:
        raw_entity (dict): Dictionary defining the raw entity detected by CM.
        entities_by_category (dict): Dictionary of previously detected entities by category.
        seen_names (set): Set of names of already detected entities.

    Returns:
        tuple: First is the updated dictionary with entities by category. Second element is the set of seen names.

    """
    entity = _reformat_cm_entity(raw_entity)

    if entity["Text"] in seen_names:
        return entities_by_category, seen_names

    entity_category, entity_type = entity["Category"], entity["Type"]

    if entity_category == "MEDICATION" and entity_type == "DURATION":
        _update_time_category_with_cm_record(
            entity, entities_by_category["TIMEDAYS"], seen_names
        )

    elif entity_category == "PROTECTED_HEALTH_INFORMATION" and entity_type == "DATE":
        _update_category_with_cm_record(
            entity, entities_by_category["TIMEYEARS"], seen_names
        )

    elif entity_category == "PROTECTED_HEALTH_INFORMATION" and entity_type == "AGE":
        _update_category_with_cm_record(entity, entities_by_category["AGE"], seen_names)

    elif entity_category == "PROTECTED_HEALTH_INFORMATION" and entity_type == "ADDRESS":
        _update_category_with_cm_record(
            entity, entities_by_category["STATE"], seen_names
        )

    elif entity_category == "MEDICATION" and (
        entity_type in ("GENERIC_NAME", "BRAND_NAME")
    ):
        _update_category_with_cm_record(
            entity, entities_by_category["DRUG"], seen_names
        )

    elif entity_category == "MEDICAL_CONDITION":
        _update_category_with_cm_record(
            entity, entities_by_category["CONDITION"], seen_names
        )

    return entities_by_category, seen_names