gendesign/tradein-mvp/backend/tests/test_consent_text_frontend_sync.py

"""ЭТАП 4 B2C launch — consent-text sync guard (part D).

WHY:
    _CONSENT_TEXT_SNAPSHOT in app/api/v1/lead.py is a durable 152-ФЗ
    proof-of-consent: it must be the EXACT text a user actually saw and agreed
    to. Before this test, the only thing keeping it in sync with the real
    frontend checkbox label (LeadForm.tsx) was a code COMMENT ("Должен
    ДОСЛОВНО совпадать с чекбоксом в LeadForm.tsx"). A comment cannot fail CI
    -- a frontend copy edit could silently drift from the backend snapshot,
    and every future lead's "proof" would then misrepresent what the user
    actually saw.

WHAT:
    Extract the actual consent-checkbox label text straight out of
    LeadForm.tsx (regex, no JSX parser needed -- there is exactly one <span>
    in the file today) and assert it matches _CONSENT_TEXT_SNAPSHOT byte-for-
    byte after whitespace normalisation (JSX text nodes wrap across source
    lines; the DOM-rendered text collapses that to single spaces). If someone
    edits ONE side without the other, this test fails.

    NOTE: the NEW anonymous-estimate consent text (_ESTIMATE_CONSENT_TEXT_SNAPSHOT
    in app/services/estimator.py, ЭТАП 4 part A) has NO frontend counterpart yet
    -- the anonymous /estimate flow isn't live (rbac_guard still requires
    X-Authenticated-User on every non-public path, see app/core/rbac.py). When
    that flow ships its own consent checkbox, add a second sync test here
    mirroring this one -- do NOT rely on a comment for that pairing either.
"""

from __future__ import annotations

import os
import re
from pathlib import Path

os.environ.setdefault("DATABASE_URL", "postgresql+psycopg://test:test@localhost:5432/test")

_FRONTEND_LEAD_FORM = (
    Path(__file__).resolve().parents[2]
    / "frontend"
    / "src"
    / "components"
    / "trade-in"
    / "v2"
    / "LeadForm.tsx"
)


def _extract_span_text(tsx_source: str) -> str:
    """Pull the text content of the (single) <span>...</span> in LeadForm.tsx,
    whitespace-normalised the same way a browser collapses JSX text-node
    whitespace when rendering (multiple lines/indentation -> single spaces).
    """
    match = re.search(r"<span>\s*(.*?)\s*</span>", tsx_source, re.DOTALL)
    assert match is not None, "no <span> found in LeadForm.tsx -- consent label markup changed"
    return re.sub(r"\s+", " ", match.group(1)).strip()


def test_frontend_lead_form_exists() -> None:
    assert _FRONTEND_LEAD_FORM.is_file(), f"missing frontend file: {_FRONTEND_LEAD_FORM}"


def test_backend_consent_snapshot_matches_frontend_checkbox_label() -> None:
    """The whole point: this FAILS if lead.py._CONSENT_TEXT_SNAPSHOT and
    LeadForm.tsx's checkbox label ever diverge -- no longer just a comment."""
    from app.api.v1.lead import _CONSENT_TEXT_SNAPSHOT

    frontend_text = _extract_span_text(_FRONTEND_LEAD_FORM.read_text(encoding="utf-8"))
    backend_text = re.sub(r"\s+", " ", _CONSENT_TEXT_SNAPSHOT).strip()

    assert frontend_text == backend_text, (
        "consent text drift detected between app/api/v1/lead.py._CONSENT_TEXT_SNAPSHOT "
        "and frontend/src/components/trade-in/v2/LeadForm.tsx checkbox label -- the "
        "152-ФЗ proof-of-consent snapshot no longer matches what users actually see. "
        "Bump _CONSENT_POLICY_VERSION and update _CONSENT_TEXT_SNAPSHOT together with "
        "any frontend copy change.\n"
        f"  frontend: {frontend_text!r}\n"
        f"  backend:  {backend_text!r}"
    )


def test_extract_span_text_helper_is_whitespace_insensitive() -> None:
    """Sanity check on the extraction helper itself, independent of the real file."""
    sample = """
        <span>
          Line one
          Line two
        </span>
    """
    assert _extract_span_text(sample) == "Line one Line two"