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

"""Tests für Cookie-Policy-Agent v3 (Sprint 1.12 Phase 2)."""

from __future__ import annotations

import asyncio

import pytest

from compliance.services.specialist_agents import (
    REGISTRY,
    AgentInput,
    CookiePolicyAgent,
    Severity,
)


FULL_POLICY = """Cookie-Richtlinie

Stand: 1. Juni 2026

Wir verwenden auf unserer Website verschiedene Cookies. Diese werden
in folgende Kategorien eingeteilt:

1. Essentielle Cookies (unbedingt erforderlich)
   Zweck: grundlegende Funktion der Website.
   Rechtsgrundlage: § 25 Abs. 2 TDDDG
   Laufzeit: Session

2. Funktionale Cookies — Zweck: Präferenzen speichern. Laufzeit: 30 Tage

3. Analytics-Cookies — Drittanbieter: Google LLC, USA
   Cookies: _ga, _gid · Laufzeit: 24 Monate
   Drittland: USA — Standardvertragsklauseln + DPF

4. Marketing — Drittanbieter: Meta Platforms Inc.
   Cookies: _fbp, _fbc · Laufzeit: 90 Tage

Cookie-Einstellungen jederzeit ändern.
Browser-Einstellungen: Chrome, Firefox, Safari, Edge.
Kontakt: datenschutz@example.com
Datenschutzbeauftragter: Max Mustermann
"""


def _run(coro):
    return asyncio.get_event_loop().run_until_complete(coro)


@pytest.fixture
def mock_v3_pipeline(monkeypatch):
    """Mockt run_v3_pipeline für deterministische Tests offline."""
    async def _fake(text, scope):
        results = [
            {"control_id": "COOKIE-MC-001",
             "passed": True, "severity": "MEDIUM",
             "label": "Cookie-Kategorien benannt",
             "regulation": "TDDDG", "article": "§ 25",
             "hint": "", "matched_text": "essentiell", "source": "kw"},
            {"control_id": "COOKIE-MC-002",
             "passed": False, "severity": "HIGH",
             "label": "Versionsdatum / Stand der Policy",
             "regulation": "DSGVO", "article": "Art. 5",
             "hint": "Bitte 'Stand: TT.MM.JJJJ' angeben",
             "matched_text": "", "source": ""},
        ]
        telemetry = {
            "layer_0_field_hits": 4,
            "layer_0_field_ids": ["categories_named", "purpose_described",
                                    "retention_duration", "version_date"],
            "layer_1_pass": 1,
            "layer_0_boost_overrides": 0,
            "total_mcs": 2,
        }
        return results, telemetry
    monkeypatch.setattr(
        "compliance.services.specialist_agents.cookie_policy.agent.run_v3_pipeline",
        _fake,
    )
    async def _no_validator(*a, **kw): return {}
    monkeypatch.setattr(
        "compliance.services.specialist_agents.cookie_policy.agent.validate_present",
        _no_validator,
    )


def test_agent_is_registered():
    a = REGISTRY.get("cookie_policy")
    assert a is not None
    assert a.doc_type == "cookie"
    assert a.agent_version == "3.0"


def test_short_text_skipped(mock_v3_pipeline):
    agent = CookiePolicyAgent()
    out = _run(agent.evaluate(AgentInput(doc_type="cookie", text="x")))
    assert all(c.status == "skipped" for c in out.mc_coverage)
    assert not out.findings


def test_agent_uses_db_mcs(mock_v3_pipeline):
    agent = CookiePolicyAgent()
    out = _run(agent.evaluate(AgentInput(doc_type="cookie",
                                            text=FULL_POLICY)))
    db_findings = [f for f in out.findings
                   if f.check_id.startswith("DBMC-")]
    assert len(db_findings) == 1
    assert db_findings[0].check_id == "DBMC-COOKIE-MC-002"
    assert db_findings[0].severity == Severity.HIGH.value


def test_agent_emits_boost_coverage(mock_v3_pipeline):
    agent = CookiePolicyAgent()
    out = _run(agent.evaluate(AgentInput(doc_type="cookie",
                                            text=FULL_POLICY)))
    # 2 DB-MCs + 12 Pattern-Boost-Slots = 14 coverage entries minimum
    assert out.mc_total >= 14
    boost_ok = [c for c in out.mc_coverage
                if c.mc_id.startswith("CP-MC-") and c.status == "ok"]
    assert len(boost_ok) == 4


def test_agent_notes_telemetry(mock_v3_pipeline):
    agent = CookiePolicyAgent()
    out = _run(agent.evaluate(AgentInput(doc_type="cookie",
                                            text=FULL_POLICY)))
    assert "v3-pipeline" in out.notes
    assert "Pattern-Boosts" in out.notes


def test_cmp_vendor_cross_check_emits_finding(mock_v3_pipeline):
    """KB-Layer: CMP-Vendoren-Cross-Check bleibt erhalten in v3."""
    agent = CookiePolicyAgent()
    out = _run(agent.evaluate(AgentInput(
        doc_type="cookie", text=FULL_POLICY,
        context={"cmp_vendors": [
            {"name": "Hotjar"},          # nicht in Policy
            {"name": "Google LLC"},       # in Policy
        ]},
    )))
    field_ids = {f.field_id for f in out.findings}
    assert "vendor_consistency" in field_ids
    f = next(f for f in out.findings
             if f.field_id == "vendor_consistency")
    assert "Hotjar" in f.evidence