breakpilot-compliance/backend-compliance/tests/test_elli_gt_coverage.py

"""GT-Coverage-Integration-Test gegen den Elli-Audit (2026-06-06).

Charakterisierungstest für die 5 GT-Lücken die in diesem Sprint
geschlossen wurden. Konstruiert einen synthetischen `state`-dict, der
Elli's reale Befunde widerspiegelt, und stellt sicher, dass jeder
der neuen B13-B16 + B9-Cleanup Detektoren genau die erwarteten
GT-IDs fängt.

Ground-Truth-Quelle: zeroclaw/docs/ground-truth/elli_eco_2026-06-06.json
"""

from compliance.services.ai_legal_basis_check import check_ai_legal_basis
from compliance.services.impressum_multi_entity_check import (
    check_multi_entity_impressum,
)
from compliance.services.retention_conflict_check import (
    check_retention_conflicts,
)
from compliance.services.url_slug_drift_check import check_url_slug_drift
from compliance.services.widerrufsbelehrung_reachability_check import (
    check_widerrufsbelehrung_reachability,
)


# Elli-Fixture: minimaler State mit den Datenfeldern die jeder Check
# braucht. Texte sind verdichtet, enthalten aber jedes prüf-relevante
# Signal (Vertex AI / lit. f, Logfile-Widerspruch, B2C-Shop, etc.).
_ELLI_IMPRESSUM = """
Impressum

Volkswagen Group Charging GmbH
Karl-Liebknecht-Str. 32, 10178 Berlin
Amtsgericht Charlottenburg HRB 208967 B
Geschäftsführer: Giovanni Palazzo, Mark Möller
Telefon: 00800 3554 1111

Elli Mobility GmbH
Karl-Liebknecht-Str. 32, 10178 Berlin
Amtsgericht Charlottenburg HRB 274616 B
USt-IdNr.: DE814424009
Geschäftsführer: Joschi Jennermann, Sebastian Steffen
"""

_ELLI_DSE = """
Datenschutzerklärung der Elli Mobility GmbH.

Server-Logfiles: Bei jedem Zugriff werden Logfiles gespeichert.
Die Logfiles werden für 7 Tage gespeichert. Bei Sicherheitsvorfällen
werden Logfiles bis zu 30 Tage aufbewahrt.

AI Assistant: Für unseren Chatbot setzen wir Google Vertex AI ein.
Rechtsgrundlage ist Art. 6 Abs. 1 lit. f DSGVO (berechtigtes
Interesse). Speicherdauer 6 Monate.

Newsletter-Abonnement: Wir speichern Anmeldedaten 24 Monate ab
Abmeldung.
"""

_ELLI_HOMEPAGE = (
    "Elli — Charging Solutions. In den Warenkorb. Lieferzeit 2 Tage. "
    "Preis inkl. MwSt. Wallbox. Naturstrom. Ladetarif buchen."
)


def _build_elli_state(widerruf_reachable: bool = False) -> dict:
    doc_entries = [
        {"doc_type": "impressum",
         "url": "https://www.elli.eco/de/impressum",
         "text": _ELLI_IMPRESSUM,
         "auto_discovered": True,
         "discovery_attempted": True},
        {"doc_type": "dse",
         "url": "https://www.elli.eco/de/datenschutz",
         "text": _ELLI_DSE,
         "auto_discovered": True,
         "discovery_attempted": True},
        {"doc_type": "cookie",
         "url": "https://www.elli.eco/de/cookies",
         "text": "x" * 600,
         "auto_discovered": True,
         "discovery_attempted": True},
        {"doc_type": "widerruf",
         "url": "https://www.elli.eco/de/widerruf" if widerruf_reachable
                else "",
         "text": ("x" * 600) if widerruf_reachable else "",
         "auto_discovered": widerruf_reachable,
         "discovery_attempted": True},
    ]
    return {
        "doc_entries": doc_entries,
        "doc_texts": {
            "impressum": _ELLI_IMPRESSUM,
            "dse": _ELLI_DSE,
        },
        "home_text": _ELLI_HOMEPAGE,
    }


# ────────────────────────────────────────────────────────────────────────
# B9 cleanup — IMPRESSUM-001
# ────────────────────────────────────────────────────────────────────────

class TestImpressum001Coverage:
    def test_per_entity_ust_id_detected(self):
        state = _build_elli_state()
        findings = check_multi_entity_impressum(state)
        ust = [f for f in findings
               if f["check_id"] == "IMPRESSUM-MULTI-UST_ID"]
        assert len(ust) == 1
        assert ust[0]["entities_missing"] == [
            "Volkswagen Group Charging GmbH",
        ]
        assert ust[0]["entities_present"] == ["Elli Mobility GmbH"]


# ────────────────────────────────────────────────────────────────────────
# B13 — WIDERRUFSBELEHRUNG-001
# ────────────────────────────────────────────────────────────────────────

class TestWiderrufsbelehrung001Coverage:
    def test_b2c_strong_high_finding(self):
        state = _build_elli_state(widerruf_reachable=False)
        findings = check_widerrufsbelehrung_reachability(state)
        assert len(findings) == 1
        assert findings[0]["check_id"] == "WIDERRUF-REACH-001"
        assert findings[0]["severity"] == "HIGH"
        assert findings[0]["b2c_scope"] == "b2c_strong"

    def test_widerruf_reachable_suppresses_finding(self):
        state = _build_elli_state(widerruf_reachable=True)
        assert check_widerrufsbelehrung_reachability(state) == []


# ────────────────────────────────────────────────────────────────────────
# B14 — TH-RETENTION-001
# ────────────────────────────────────────────────────────────────────────

class TestTHRetention001Coverage:
    def test_logfile_7_vs_30_finding(self):
        state = _build_elli_state()
        findings = check_retention_conflicts(state)
        log = [f for f in findings if f["category"] == "logfile"]
        assert len(log) == 1
        assert 7.0 in log[0]["values_days"]
        assert 30.0 in log[0]["values_days"]


# ────────────────────────────────────────────────────────────────────────
# B15 — AI-ACT-RISK-001
# ────────────────────────────────────────────────────────────────────────

class TestAIActRisk001Coverage:
    def test_vertex_ai_lit_f_finding(self):
        state = _build_elli_state()
        findings = check_ai_legal_basis(state)
        assert len(findings) == 1
        assert findings[0]["check_id"] == "AI-LEGAL-BASIS-001"
        assert "Vertex" in findings[0]["provider"]


# ────────────────────────────────────────────────────────────────────────
# B16 — URL-STRUCTURE-001
# ────────────────────────────────────────────────────────────────────────

class TestURLStructure001Coverage:
    def test_cookie_richtlinie_404_emits_finding(self):
        # Mock _head_status: alle Standard-Slugs ausser dem discovered
        # liefern 404 — Elli's reale Lage.
        from unittest.mock import patch
        state = _build_elli_state()
        with patch(
            "compliance.services.url_slug_drift_check._head_status",
            return_value=404,
        ):
            findings = check_url_slug_drift(state)
        cookie_findings = [f for f in findings if f["doc_type"] == "cookie"]
        assert len(cookie_findings) == 1
        assert cookie_findings[0]["severity"] == "LOW"
        assert "cookie-richtlinie" in cookie_findings[0]["alt_slugs_404"]


# ────────────────────────────────────────────────────────────────────────
# All-Detector-Smoke — sicherstellen, dass die 5 Checks zusammen
# gegen Elli mindestens 5 unterschiedliche Finding-IDs ergeben.
# ────────────────────────────────────────────────────────────────────────

class TestElliFullCoverage:
    def test_all_five_detectors_fire(self):
        from unittest.mock import patch
        state = _build_elli_state(widerruf_reachable=False)
        all_findings: list[dict] = []
        all_findings.extend(check_multi_entity_impressum(state))
        all_findings.extend(check_widerrufsbelehrung_reachability(state))
        all_findings.extend(check_retention_conflicts(state))
        all_findings.extend(check_ai_legal_basis(state))
        with patch(
            "compliance.services.url_slug_drift_check._head_status",
            return_value=404,
        ):
            all_findings.extend(check_url_slug_drift(state))
        check_ids = {f["check_id"] for f in all_findings}
        expected = {
            "IMPRESSUM-MULTI-UST_ID",
            "WIDERRUF-REACH-001",
            "RETENTION-CONFLICT-001",
            "AI-LEGAL-BASIS-001",
            "URL-SLUG-DRIFT-001",
        }
        missing = expected - check_ids
        assert not missing, f"GT-Lücken nicht gefangen: {missing}"