Files
breakpilot-compliance/consent-tester/tests/test_cookieless_optout.py
T
Benjamin Admin 11740bd2f9 feat(consent-tester): 4 weitere Edge-Cases — Consent-or-Pay, Consent Mode, CNAME-Cloaking, Returning-User
#4 Consent-or-Pay (EDPB Opinion 08/2024): Banner-Text-Signatur (Pur-Abo/
   "zustimmen oder bezahlen" + Consent-Kontext) → MEDIUM-Befund "rechtlich
   umstritten, gesondert prüfen".
#5 Google Consent Mode v2: page.evaluate (dataLayer-consent-Events / inline
   gtag('consent')) → MEDIUM "ist KEINE gültige Einwilligung".
#6 CNAME-Cloaking: First-Party-Subdomains per socket.gethostbyname_ex auflösen,
   CNAME-Kette gegen bekannte Tracker-Infra (Eulerian/Adobe/Webtrekk/…) → HIGH
   "faktisch Drittanbieter trotz First-Party-Optik". Best-effort, kurze Timeouts.
#7 Returning-User: Scanner nutzt by-design frische Browser-Contexts → Hinweis im
   Kein-Banner-Befund (fehlendes Banner liegt nicht an erinnertem Consent).

Tests + py_compile grün.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-12 20:45:20 +02:00

96 lines
3.7 KiB
Python

"""Cookieless Opt-out-Erkennung im banner_text_checker.
Sehr untypischer Sonderfall: cookie-FREIE Analyse mit reinem Opt-out-Hinweis
statt Consent-Banner (z.B. bayshore.ai). Standard-Opt-in-Checks duerfen dann
NICHT feuern (sonst False Positives).
"""
from services.banner_text_checker import (
is_cookieless_optout,
detect_non_cookie_tracking,
build_no_banner_finding,
detect_consent_or_pay,
detect_cname_cloaking,
)
def test_bayshore_cookieless_optout_detected():
# Realer bayshore.ai-Bannertext (Opt-out fuer cookie-freie Analyse).
bay = ("privacy-friendly, cookie-free analytics are currently enabled. "
"you can change your choice at any time. disable")
assert is_cookieless_optout(bay) is True
def test_standard_consent_banner_not_cookieless():
assert not is_cookieless_optout(
"wir nutzen cookies. alle akzeptieren ablehnen einstellungen")
def test_cookiefree_but_with_accept_is_not_optout():
# 'cookie-free' genannt, aber echtes Consent ('accept all') → kein reiner Opt-out.
assert not is_cookieless_optout("cookie-free analytics. accept all disable")
def test_signal_without_optout_word_is_not_detected():
# cookie-free, aber kein Opt-out-Mechanismus im Text.
assert not is_cookieless_optout("cookie-free analytics enabled")
def test_empty():
assert not is_cookieless_optout("")
# ── #3: Non-Cookie-Tracking-Erkennung ──────────────────────────────────
def test_detect_meta_pixel():
assert detect_non_cookie_tracking(
["https://connect.facebook.net/en_US/fbevents.js"]) == ["Meta-Pixel (Facebook)"]
def test_detect_clarity_and_fingerprint():
found = detect_non_cookie_tracking([
"https://www.clarity.ms/tag/abc", "https://cdn.fpjs.io/v3/x.js"])
assert "Microsoft Clarity" in found
assert "Fingerprinting (FingerprintJS)" in found
def test_detect_none_on_plain_scripts():
assert detect_non_cookie_tracking(
["https://example.com/app.js", "/static/main.css"]) == []
assert detect_non_cookie_tracking([]) == []
# ── #1/#2: Kein-Banner-affirmativ-Befund ───────────────────────────────
def test_no_banner_finding_is_low_and_compliant():
v = build_no_banner_finding(has_dse=True)
assert v.severity == "LOW"
assert "konform" in v.text.lower()
assert "geo-targeting" in v.text.lower() # Geo-Caveat enthalten
def test_no_banner_finding_flags_missing_dse():
v = build_no_banner_finding(has_dse=False)
assert "dse" in v.text.lower() or "datenschutzerkl" in v.text.lower()
# ── #4: Consent-or-Pay ─────────────────────────────────────────────────
def test_consent_or_pay_detected():
assert detect_consent_or_pay(
"akzeptieren oder pur-abo abschliessen. cookies & werbung") is True
def test_consent_or_pay_not_on_standard_banner():
assert not detect_consent_or_pay("alle akzeptieren ablehnen einstellungen cookies")
def test_consent_or_pay_needs_consent_context():
# 'Abo' ohne Consent-Kontext (z.B. normale Paywall) ist kein Consent-or-Pay.
assert not detect_consent_or_pay("jetzt abonnieren fuer 5 pro monat")
# ── #6: CNAME-Cloaking (nur DNS-freie Faelle) ──────────────────────────
def test_cname_cloaking_empty_and_no_subdomain():
assert detect_cname_cloaking([], "example.com") == []
assert detect_cname_cloaking(["https://example.com/app.js"], "") == []
# Script auf der Hauptdomain (keine Subdomain) → kein Lookup → []
assert detect_cname_cloaking(["https://example.com/app.js"], "example.com") == []