Function normalize_label

openmed/core/labels.py:670–708

Normalize an entity label to the canonical taxonomy. Accepts any of: - English lowercase ``snake_case`` (``first_name``) - Portuguese ``UPPERCASE`` no-separator (``FIRSTNAME``) - BIOES-tagged forms (``B-NAME``, ``I-EMAIL``) - Mixed case with arbitrary separators (``First

(label: str, lang: str = "en")

def normalize_label(label: str, lang: str = "en") -> str:
    """Normalize an entity label to the canonical taxonomy.

    Accepts any of:
      - English lowercase ``snake_case`` (``first_name``)
      - Portuguese ``UPPERCASE`` no-separator (``FIRSTNAME``)
      - BIOES-tagged forms (``B-NAME``, ``I-EMAIL``)
      - Mixed case with arbitrary separators (``First-Name``, ``First Name``)

    Unknown labels fall through to ``OTHER`` rather than raising — callers
    that need strict checking should compare against ``CANONICAL_LABELS``
    explicitly.

    Args:
        label: Source label as emitted by a model or registered in a config.
        lang: ISO 639-1 language hint (currently unused but reserved for
            language-conditional disambiguation, e.g. mapping ambiguous
            tokens differently per locale).

    Returns:
        A canonical label in ``UPPER_SNAKE_CASE``.
    """
    if not label:
        return OTHER
    key = _key(label)
    if not key:
        return OTHER
    canonical = _ALIAS_MAP.get(key)
    if canonical is not None:
        return canonical
    # If the input already matches a canonical label after stripping
    # separators (e.g. ``ID_NUM`` -> key ``idnum`` -> aliased; but
    # ``CREDIT_CARD`` -> ``creditcard`` -> aliased), the alias map covers
    # it. The ``upper`` fallback handles any future canonical label not
    # yet in the alias map.
    upper = re.sub(r"[^A-Z0-9_]", "", label.upper().replace("-", "_").replace(" ", "_"))
    if upper in CANONICAL_LABELS:
        return upper
    return OTHER


def id_subtype_for(label: str, lang: str = "en") -> str | None: