breakpilot-compliance/consent-tester/checks/banner_runner.py

"""
Banner Runner — maps scan results to the L1/L2 check hierarchy.

Takes the raw ScanResponse dict and produces a structured_checks list
compatible with ChecklistView (same format as document checks).
"""

from checks.banner_checks import BANNER_CHECKLIST


def map_scan_to_checks(scan_result: dict) -> dict:
    """Map a /scan response to the L1/L2 banner check hierarchy.

    Returns dict with:
      - structured_checks: list of CheckItem dicts
      - completeness_pct: L1 pass rate (0-100)
      - correctness_pct: L2 pass rate (0-100)
    """
    # Collect all violation codes from every source
    violation_codes = _collect_violation_codes(scan_result)

    # Collect pass codes — some checks produce boolean signals, not violations
    pass_codes = _collect_pass_codes(scan_result)

    # Build structured checks
    checks: list[dict] = []
    l1_checks: list[dict] = []
    l2_checks: list[dict] = []

    for defn in BANNER_CHECKLIST:
        key = defn["check_key"]
        level = defn["level"]
        parent = defn.get("parent")

        # Determine pass/fail
        is_violation_key = key in violation_codes
        is_pass_key = key in pass_codes

        # For checks whose check_key appears in violations → failed
        # For checks whose check_key appears only in passes → passed
        # For checks where neither:
        #   - Phase-based checks (tracking/cookies) → PASS (absence = good)
        #   - Banner UI checks → PASS only if banner was detected and
        #     the scanner actually ran the relevant check
        if is_violation_key:
            passed = False
            matched_text = violation_codes[key]
        elif is_pass_key:
            passed = True
            matched_text = pass_codes.get(key, "")
        else:
            banner_detected = scan_result.get("banner_detected", False)
            if key == "banner_detected":
                passed = banner_detected
            elif key in _ABSENCE_IS_PASS:
                # For these checks, no violation = passed (e.g. no tracking cookies)
                passed = True
            elif banner_detected:
                # Banner was detected but this specific check produced no result.
                # If the scanner ran banner_checks → assume checked and passed.
                # If banner_checks is empty → scanner couldn't test → not passed.
                has_banner_results = bool(scan_result.get("banner_checks", {}).get("violations") is not None)
                passed = has_banner_results
            else:
                passed = False
            matched_text = ""

        # L2 checks are skipped if their parent L1 failed
        skipped = False
        if level == 2 and parent:
            parent_check = next(
                (c for c in checks if c["id"] == parent), None
            )
            if parent_check and not parent_check["passed"]:
                skipped = True

        item = {
            "id": defn["id"],
            "label": defn["label"],
            "passed": passed and not skipped,
            "severity": defn["severity"],
            "level": level,
            "parent": parent,
            "skipped": skipped,
            "hint": defn.get("hint", ""),
            "matched_text": matched_text if passed else "",
        }
        checks.append(item)

        if level == 1:
            l1_checks.append(item)
        elif level == 2:
            l2_checks.append(item)

    # Compute percentages
    l1_total = len(l1_checks)
    l1_passed = sum(1 for c in l1_checks if c["passed"])
    completeness_pct = round(l1_passed / l1_total * 100) if l1_total else 0

    l2_active = [c for c in l2_checks if not c["skipped"]]
    l2_passed = sum(1 for c in l2_active if c["passed"])
    correctness_pct = round(l2_passed / len(l2_active) * 100) if l2_active else 0

    return {
        "structured_checks": checks,
        "completeness_pct": completeness_pct,
        "correctness_pct": correctness_pct,
    }


_TEXT_TO_CODE: list[tuple[str, str]] = [
    ("impressum", "impressum_link"),
    ("erneuter zugang", "re_access_settings"),
    ("cookie-einstellung", "re_access_settings"),
    ("widerruf der einwilligung", "re_access_settings"),
    ("vorausgewaehlte", "pre_ticked_checkboxes"),
    ("vorausgew", "pre_ticked_checkboxes"),
    ("akzeptieren.*groesser", "dark_pattern_button_size"),
    ("akzeptieren.*gr\u00f6\u00dfer", "dark_pattern_button_size"),
    ("hintergrundfarbe", "color_contrast_dark_pattern"),
    ("optisch kaum sichtbar", "color_contrast_dark_pattern"),
    ("dark pattern", "color_contrast_dark_pattern"),
    ("cookie wall", "cookie_wall"),
    ("ablehnen.*button", "reject_button_visible"),
    ("kein sichtbarer", "reject_button_visible"),
    ("zustimmung zur datenschutz", "wrong_dse_consent"),
    ("consent mode", "google_consent_mode_defaults"),
    ("tracking.*vor consent", "cookies_before_consent"),
    ("tracking-cookie", "cookies_before_consent"),
    ("nicht modal", "non_modal_dismiss"),
    ("hintergrund.*schliessen", "non_modal_dismiss"),
    ("klick.*asymmetri", "click_count_asymmetry"),
    ("ablehnung.*klick", "click_count_asymmetry"),
    ("koppelungsverbot", "registration_consent_coupling"),
    ("registrierung", "registration_consent_coupling"),
    ("sprache.*stimmt nicht", "banner_language_mismatch"),
    ("banner-sprache", "banner_language_mismatch"),
    ("consent-cookie.*laeuft", "consent_cookie_expiry_13m"),
    ("consent-cookie.*l\u00e4uft", "consent_cookie_expiry_13m"),
    ("13 monate", "consent_cookie_expiry_13m"),
    ("nudging", "nudging_reject_hidden"),
    ("scrollen", "nudging_reject_hidden"),
    ("emotionale sprache", "stirring_emotional_language"),
    ("stirring", "stirring_emotional_language"),
    ("drittanbieter.*dse", "third_party_dse_link"),
    ("ohne vorherige einwilligung", "tracking_before_consent"),
    ("trotz ablehnung", "tracking_after_reject"),
]


def _text_to_code(text: str) -> str:
    """Infer a check_key from violation text content."""
    t = text.lower()
    for pattern, code in _TEXT_TO_CODE:
        if pattern in t:
            return code
    return ""


def _collect_violation_codes(scan: dict) -> dict[str, str]:
    """Collect check_key → violation text from all sources."""
    codes: dict[str, str] = {}

    # Banner text violations — match by code field OR by text content
    banner_checks = scan.get("banner_checks", {})
    for v in banner_checks.get("violations", []):
        code = v.get("code", "") or _text_to_code(v.get("text", ""))
        if code:
            codes[code] = v.get("text", "")[:120]

    # Phase A violations (before consent)
    phase_a = scan.get("phases", {}).get("before_consent", {})
    for v in phase_a.get("violations", []):
        code = v.get("code", "") or _text_to_code(v.get("text", ""))
        if code:
            codes[code] = v.get("text", "")[:120]

    # Phase B violations (after reject)
    phase_b = scan.get("phases", {}).get("after_reject", {})
    for v in phase_b.get("violations", []):
        code = v.get("code", "") or _text_to_code(v.get("text", ""))
        if code:
            codes[code] = v.get("text", "")[:120]

    # Tracking services in phase A → tracking_before_consent
    tracking_a = phase_a.get("tracking_services", [])
    if tracking_a and "tracking_before_consent" not in codes:
        codes["tracking_before_consent"] = ", ".join(tracking_a[:5])

    # Cookies before consent → cookies_before_consent
    cookies_a = phase_a.get("cookies", [])
    tracking_cookies = [c for c in cookies_a if _is_tracking_cookie(c)]
    if tracking_cookies and "cookies_before_consent" not in codes:
        codes["cookies_before_consent"] = ", ".join(tracking_cookies[:5])

    # New tracking after reject → tracking_after_reject
    new_tracking_b = phase_b.get("new_tracking", [])
    if new_tracking_b and "tracking_after_reject" not in codes:
        codes["tracking_after_reject"] = ", ".join(new_tracking_b[:5])

    return codes


def _collect_pass_codes(scan: dict) -> dict[str, str]:
    """Collect explicit pass signals from scan results."""
    passes: dict[str, str] = {}

    # Banner detected
    if scan.get("banner_detected"):
        passes["banner_detected"] = scan.get("banner_provider", "detected")

    # Provider named
    provider = scan.get("banner_provider", "")
    if provider:
        passes["banner_provider_named"] = provider

    # Impressum link
    bc = scan.get("banner_checks", {})
    if bc.get("has_impressum_link"):
        passes["impressum_link"] = "Impressum-Link gefunden"
    if bc.get("has_dse_link"):
        passes["dse_link"] = "DSE-Link gefunden"

    return passes


# Checks where absence of a violation means PASS (not "untested")
# These are phase-based checks: if no tracking was detected, that's good.
_ABSENCE_IS_PASS = {
    "tracking_before_consent",
    "cookies_before_consent",
    "tracking_after_reject",
    "google_consent_mode_defaults",
    "banner_language_mismatch",
    "cookie_wall",
}

_TRACKING_COOKIE_PREFIXES = (
    "_ga", "_gid", "_fbp", "_fbc", "IDE", "_gcl", "fr", "_pin",
    "_tt_", "li_sugr", "_hj", "mp_", "ajs_", "_clck", "_clsk",
)


def _is_tracking_cookie(name: str) -> bool:
    """Check if a cookie name is a known tracking cookie."""
    return any(name.startswith(p) for p in _TRACKING_COOKIE_PREFIXES)