from __future__ import annotations

from collections.abc import Iterable

# The fixed v0 governance classification set (ADR 0011). This is a platform invariant,
# not source config: source config decides *which* of these to apply to a table, but the
# valid universe is fixed here so a typo or an arbitrary label in source.yaml cannot flow
# into governance metadata. Keep this in sync with
# docs/adr/0011-v0-starts-with-a-small-governance-classification-set.md.
GOVERNANCE_CLASSIFICATIONS: frozenset[str] = frozenset(
    {
        "public",
        "internal",
        "confidential",
        "restricted",
        "pii",
        "financial",
        "customer_data",
        "operational_sensitive",
    }
)


def validate_classifications(classifications: Iterable[str], *, context: str = "") -> list[str]:
    """Return the classifications as a list, raising if any is outside the ADR-0011 set.

    ``context`` is included in the error message to point at the offending source
    (e.g. ``"table account"``).
    """
    result = list(classifications)
    unknown = sorted({c for c in result if c not in GOVERNANCE_CLASSIFICATIONS})
    if unknown:
        where = f" ({context})" if context else ""
        raise ValueError(
            f"unknown governance classification(s){where}: {unknown}; "
            f"allowed set (ADR 0011): {sorted(GOVERNANCE_CLASSIFICATIONS)}"
        )
    return result