backend/tests/test_pii_whitelist_source.py

"""Guard for CONSTANT 2.1 relocation (see backend/docs/RICS_CONSTANT_INVENTORY.md).

``pii_scrubber._STATUS_TERMS`` previously inlined the RICS rating label
"condition rating". Post-refactor the rating term is derived from the canonical
``DEFAULT_RATING_SYSTEM_NAME`` (rics_canonical_l3.py) and the remaining terms
live in ``_BASE_STATUS_TERMS``. The effective whitelist must stay byte-identical
as a set — zero terms added or removed — and the relocated term must still be
protected from redaction.

Pre-refactor this module FAILS at import (``_BASE_STATUS_TERMS`` does not exist);
post-refactor it PASSES.
"""

from __future__ import annotations

from backend.core import pii_scrubber
from backend.core.pii_scrubber import (
    _BASE_STATUS_TERMS,
    _STATUS_TERMS,
    PROPTECH_SAFE_WHITELIST,
)
from backend.core.rics_canonical_l3 import DEFAULT_RATING_SYSTEM_NAME

# Exactly the terms enumerated in the inventory for 2.1 — no more, no less.
_EXPECTED_STATUS_TERMS = frozenset(
    {
        "condition rating",
        "defect",
        "satisfactory",
        "serviceable",
        "deflection",
        "distortion",
        "cracking",
        "spalled",
        "damp",
        "moisture",
        "insulation",
    }
)


def test_status_terms_exactly_match_inventory():
    assert _STATUS_TERMS == _EXPECTED_STATUS_TERMS  # no terms added or removed


def test_rating_term_is_canonical_derived_not_inlined():
    rating_term = DEFAULT_RATING_SYSTEM_NAME.lower()
    assert rating_term == "condition rating"
    assert rating_term in _STATUS_TERMS
    # The rating term must come from the canonical seed, not the inlined base set.
    assert rating_term not in _BASE_STATUS_TERMS
    assert _STATUS_TERMS == _BASE_STATUS_TERMS | frozenset({rating_term})


def test_status_terms_remain_in_global_whitelist():
    assert _STATUS_TERMS <= PROPTECH_SAFE_WHITELIST


def test_relocated_rating_phrase_still_survives_scrub():
    # PII boundary must not shift: the protected phrase stays unredacted.
    text = "The survey records a condition rating against the roof covering."
    result = pii_scrubber.scrub(text)
    assert "condition rating" in result.text.lower()
    assert result.total == 0