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). The label
    now wraps a `<Link>` ("Политикой обработки персональных данных" is a
    clickable link to the actual policy document, RKN/owner requirement --
    the extractor strips JSX tags AND `{" "}` expression-spacers, keeping only
    the human-readable text, so the comparison stays a FLAT string on both
    sides). If someone edits ONE side without the other, this test fails.

    A second test (`test_consent_policy_version_matches_privacy_approval_date`)
    guards the OTHER half of the same drift class found during triage: nothing
    was checking that _CONSENT_POLICY_VERSION actually points at the privacy
    policy edition it claims to (PRIVACY_APPROVAL in mera-public/content.ts).
    Bumping the policy text without bumping the version tag (or vice versa)
    would silently mislabel every lead's proof-of-consent snapshot.

    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")

_REPO_ROOT = Path(__file__).resolve().parents[2]

_FRONTEND_LEAD_FORM = (
    _REPO_ROOT / "frontend" / "src" / "components" / "trade-in" / "v2" / "LeadForm.tsx"
)

_FRONTEND_LEGAL_CONTENT = _REPO_ROOT / "frontend" / "src" / "app" / "mera-public" / "content.ts"

# Родительный падеж месяцев, как их пишет владелец в content.ts ("13 августа 2026 г.").
_RU_MONTHS_GENITIVE = {
    "января": 1,
    "февраля": 2,
    "марта": 3,
    "апреля": 4,
    "мая": 5,
    "июня": 6,
    "июля": 7,
    "августа": 8,
    "сентября": 9,
    "октября": 10,
    "ноября": 11,
    "декабря": 12,
}


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).

    The span may contain nested JSX markup (e.g. a <Link> wrapping part of the
    label, and a `{" "}` expression-spacer forcing a real space between a text
    node and the link on the next source line -- plain JSX whitespace between
    a text node and a tag on separate lines collapses to NOTHING, not a
    space, so LeadForm.tsx needs that explicit spacer for correct rendering).
    Both are stripped here so the comparison is against the flat, human-
    readable text a user actually sees -- not the markup.
    """
    match = re.search(r"<span>(.*?)</span>", tsx_source, re.DOTALL)
    assert match is not None, "no <span> found in LeadForm.tsx -- consent label markup changed"
    inner = match.group(1)
    inner = re.sub(r"\{\s*[\"']\s*[\"']\s*\}", " ", inner)  # {" "} spacer -> real space
    inner = re.sub(r"\{/\*.*?\*/\}", " ", inner, flags=re.DOTALL)  # JSX comments
    inner = re.sub(r"<[^>]+>", "", inner)  # strip remaining JSX tags (e.g. <Link ...>, </Link>)
    return re.sub(r"\s+", " ", inner).strip()


def _extract_privacy_approval_iso_date(content_ts_source: str) -> str:
    """Pull the "DD <month genitive> YYYY" date out of PRIVACY_APPROVAL in
    mera-public/content.ts and return it as an ISO "YYYY-MM-DD" string.

    PRIVACY_APPROVAL ("приказом директора № 1 от 13 августа 2026 г.") is the
    order that approves the actual privacy-policy EDITION the consent
    checkbox links to (/mera-public/privacy) -- it is the correct source of
    truth for _CONSENT_POLICY_VERSION, as opposed to LEGAL_DOCS_REVISION
    (which dates the offer + refund-policy documents, a different pair).
    """
    match = re.search(r'PRIVACY_APPROVAL\s*=\s*"([^"]+)"', content_ts_source)
    assert match is not None, "PRIVACY_APPROVAL constant not found in mera-public/content.ts"
    date_match = re.search(r"(\d{1,2})\s+([а-яё]+)\s+(\d{4})", match.group(1))
    assert date_match is not None, f"no RU date found in PRIVACY_APPROVAL: {match.group(1)!r}"
    day, month_name, year = date_match.groups()
    month = _RU_MONTHS_GENITIVE.get(month_name)
    assert month is not None, f"unknown RU month name in PRIVACY_APPROVAL: {month_name!r}"
    return f"{year}-{month:02d}-{int(day):02d}"


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_consent_policy_version_matches_privacy_approval_date() -> None:
    """Guards the other half of the same drift class as the test above:
    _CONSENT_POLICY_VERSION must point at the privacy-policy EDITION it
    claims to (PRIVACY_APPROVAL in mera-public/content.ts), not just be some
    unrelated date bumped by hand. A silent mismatch here would mislabel
    every lead's proof-of-consent snapshot with the wrong policy edition."""
    from app.api.v1.lead import _CONSENT_POLICY_VERSION

    assert _FRONTEND_LEGAL_CONTENT.is_file(), f"missing frontend file: {_FRONTEND_LEGAL_CONTENT}"
    expected_version = _extract_privacy_approval_iso_date(
        _FRONTEND_LEGAL_CONTENT.read_text(encoding="utf-8")
    )

    assert _CONSENT_POLICY_VERSION == expected_version, (
        "app/api/v1/lead.py._CONSENT_POLICY_VERSION does not match the privacy-policy "
        "edition date derived from PRIVACY_APPROVAL in frontend/src/app/mera-public/"
        "content.ts. Bump _CONSENT_POLICY_VERSION to the new edition date whenever "
        "PRIVACY_APPROVAL changes (or vice versa).\n"
        f"  _CONSENT_POLICY_VERSION: {_CONSENT_POLICY_VERSION!r}\n"
        f"  PRIVACY_APPROVAL date:   {expected_version!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"


def test_extract_span_text_helper_strips_nested_link_and_spacer() -> None:
    """Sanity check: a <Link> wrapping part of the label (plus the {" "}
    spacer JSX needs to force a real space before it) must collapse to plain
    text, exactly like a browser renders it -- this is the shape LeadForm.tsx
    actually uses today for the policy-document link."""
    sample = """
        <span>
          Согласен(-на) на обработку персональных данных в соответствии с{" "}
          <Link href={PRIVACY_PATH} target="_blank" rel="noreferrer">
            Политикой обработки персональных данных
          </Link>
        </span>
    """
    assert _extract_span_text(sample) == (
        "Согласен(-на) на обработку персональных данных в соответствии с "
        "Политикой обработки персональных данных"
    )


def test_extract_privacy_approval_iso_date_helper() -> None:
    """Sanity check on the RU-date extraction helper, independent of the real file."""
    sample = 'export const PRIVACY_APPROVAL = "приказом директора № 1 от 13 августа 2026 г.";'
    assert _extract_privacy_approval_iso_date(sample) == "2026-08-13"