app/dataops/normalizer.py

"""Entity normalizer."""

from __future__ import annotations

from app.knowledge.drug_catalog import canonicalize_drug_name


def normalize_drug_entities(items: list[str]) -> list[str]:
    return sorted({canonicalize_drug_name(item) for item in items})


def normalize_component_entities(items: list[str]) -> list[str]:
    return sorted({canonicalize_drug_name(item).replace("-", "_") for item in items if item})