152-fz-scanner/tests/test_smoke.py

"""
Smoke-тест: проверяет, что все 7 быстрых проверок импортируются,
создаются и запускаются на минимальной HTML-странице без ошибок.
"""
import pytest
from pathlib import Path
import sys

sys.path.insert(0, str(Path(__file__).parent.parent))

from scanner.checks import ALL_CHECKS, QUICK_CHECKS
from scanner.models import CheckStatus

# Минимальная HTML-страница для тестирования
MOCK_HTML = """<!DOCTYPE html>
<html>
<head><title>Test Site</title></head>
<body>
    <footer>
        <a href="/privacy">Политика конфиденциальности</a>
        <a href="/consent">Согласие на обработку персональных данных</a>
    </footer>
    <form action="/submit" method="POST">
        <input type="text" name="name" placeholder="Имя">
        <input type="email" name="email" placeholder="Email">
        <label for="agree">
            <input type="checkbox" id="agree" name="agree">
            Я даю согласие на обработку персональных данных
        </label>
        <button type="submit">Отправить</button>
    </form>
    <div id="cookie-banner" class="cookie-consent">
        <p>Мы используем cookies</p>
        <button id="accept-cookies">Принять</button>
        <button id="reject-cookies">Отказаться</button>
    </div>
    <script src="https://mc.yandex.ru/metrika/tag.js"></script>
</body>
</html>"""


class TestAllChecksImport:
    """Проверка что все проверки импортируются."""

    def test_quick_checks_count(self):
        assert len(QUICK_CHECKS) == 7, f"Expected 7 quick checks, got {len(QUICK_CHECKS)}"

    def test_all_checks_registered(self):
        for name in QUICK_CHECKS:
            assert name in ALL_CHECKS, f"Check '{name}' not in ALL_CHECKS"

    @pytest.mark.parametrize("check_name", QUICK_CHECKS)
    def test_check_instantiable(self, check_name):
        """Каждая проверка создаётся без ошибок."""
        check_cls = ALL_CHECKS[check_name]
        instance = check_cls()
        assert instance.name == check_name
        assert instance.title


class TestCheckLogic:
    """Проверка логики проверок (без Playwright — unit tests)."""

    def test_form_checkbox_patterns(self):
        from scanner.checks.form_checkbox import FormCheckboxCheck
        import re
        check = FormCheckboxCheck()
        # Паттерны должны матчить русский текст согласия
        assert any(re.search(p, "согласие на обработку данных", re.IGNORECASE) for p in check.CONSENT_PATTERNS)
        assert any(re.search(p, "я согласен на обработку персональных данных", re.IGNORECASE) for p in check.CONSENT_PATTERNS)

    def test_operator_info_inn_pattern(self):
        from scanner.checks.operator_info import OperatorInfoCheck
        import re
        check = OperatorInfoCheck()
        m = re.search(check.INN_PATTERN, "ИНН: 0278901234", re.IGNORECASE)
        assert m, "INN pattern should match 10-digit INN"
        assert m.group(1) == "0278901234"

    def test_external_scripts_yaml_loads(self):
        from scanner.checks.external_scripts import ExternalScriptsCheck
        check = ExternalScriptsCheck()
        trackers = check._load_trackers()
        assert "critical" in trackers
        assert "russian" in trackers
        assert any(e["domain"] == "google-analytics.com" for e in trackers["critical"])


class TestScoring:
    """Проверка логики скоринга."""

    def test_score_calculation_critical(self):
        from scanner.models import SiteScan, CheckResult, CheckStatus, RiskLevel
        from datetime import datetime
        scan = SiteScan(url="https://test.ru", scanned_at=datetime.now())
        scan.checks = [
            CheckResult(name="t1", title="Test 1", status=CheckStatus.FAIL, risk_level=RiskLevel.CRITICAL),
            CheckResult(name="t2", title="Test 2", status=CheckStatus.PASS, risk_level=RiskLevel.LOW),
        ]
        scan._calculate_score()
        assert scan.score == 70  # 100 - 30
        assert scan.is_compliant  # score=70 — на границе, но compliant

    def test_score_all_pass(self):
        from scanner.models import SiteScan, CheckResult, CheckStatus, RiskLevel
        from datetime import datetime
        scan = SiteScan(url="https://test.ru", scanned_at=datetime.now())
        scan.checks = [
            CheckResult(name="t1", title="Test 1", status=CheckStatus.PASS, risk_level=RiskLevel.HIGH),
        ]
        scan._calculate_score()
        assert scan.score == 100
        assert scan.is_compliant

    def test_score_multiple_critical(self):
        from scanner.models import SiteScan, CheckResult, CheckStatus, RiskLevel
        from datetime import datetime
        scan = SiteScan(url="https://test.ru", scanned_at=datetime.now())
        scan.checks = [
            CheckResult(name="t1", title="C1", status=CheckStatus.FAIL, risk_level=RiskLevel.CRITICAL),
            CheckResult(name="t2", title="C2", status=CheckStatus.FAIL, risk_level=RiskLevel.CRITICAL),
            CheckResult(name="t3", title="H1", status=CheckStatus.FAIL, risk_level=RiskLevel.HIGH),
        ]
        scan._calculate_score()
        assert scan.score == 20  # 100 - 30 - 30 - 20
        assert not scan.is_compliant


class TestEnginePhases:
    """Проверка фазовой модели в engine."""

    def test_mutating_checks_identified(self):
        """cookie_audit должен быть в mutating_checks."""
        from scanner.checks import QUICK_CHECKS
        mutating = {"cookie_audit"}
        assert "cookie_audit" in QUICK_CHECKS, "cookie_audit должен быть в QUICK_CHECKS"
        non_mutating = [n for n in QUICK_CHECKS if n not in mutating]
        assert len(non_mutating) == 6  # 7 quick - 1 mutating