breakpilot-compliance/backend-compliance/tests/test_v1_enrichment.py

"""Tests for V1 Control Enrichment (Eigenentwicklung matching)."""
import sys
sys.path.insert(0, ".")

import pytest
from unittest.mock import AsyncMock, MagicMock, patch

from compliance.services.v1_enrichment import (
    enrich_v1_matches,
    get_v1_matches,
    count_v1_controls,
)


class TestV1EnrichmentDryRun:
    """Dry-run mode should return statistics without touching DB."""

    @pytest.mark.asyncio
    async def test_dry_run_returns_stats(self):
        mock_v1 = [
            MagicMock(
                id="uuid-v1-1",
                control_id="ACC-013",
                title="Zugriffskontrolle",
                objective="Zugriff einschraenken",
                category="access",
            ),
            MagicMock(
                id="uuid-v1-2",
                control_id="SEC-005",
                title="Verschluesselung",
                objective="Daten verschluesseln",
                category="encryption",
            ),
        ]

        mock_count = MagicMock(cnt=863)

        with patch("compliance.services.v1_enrichment.SessionLocal") as mock_session:
            db = MagicMock()
            mock_session.return_value.__enter__ = MagicMock(return_value=db)
            mock_session.return_value.__exit__ = MagicMock(return_value=False)
            # First call: v1 controls, second call: count
            db.execute.return_value.fetchall.return_value = mock_v1
            db.execute.return_value.fetchone.return_value = mock_count

            result = await enrich_v1_matches(dry_run=True, batch_size=100, offset=0)

        assert result["dry_run"] is True
        assert result["total_v1"] == 863
        assert len(result["sample_controls"]) == 2
        assert result["sample_controls"][0]["control_id"] == "ACC-013"


class TestV1EnrichmentExecution:
    """Execution mode should find matches and insert them."""

    @pytest.mark.asyncio
    async def test_processes_and_inserts_matches(self):
        mock_v1 = [
            MagicMock(
                id="uuid-v1-1",
                control_id="ACC-013",
                title="Zugriffskontrolle",
                objective="Zugriff auf Systeme einschraenken",
                category="access",
            ),
        ]

        mock_count = MagicMock(cnt=1)

        # Atomic control found in Qdrant (has parent, no source_citation)
        mock_atomic_row = MagicMock(
            id="uuid-atomic-1",
            control_id="SEC-042-A01",
            title="Verschluesselung (atomar)",
            source_citation=None,  # Atomic controls don't have source_citation
            parent_control_uuid="uuid-reg-1",
            severity="high",
            category="encryption",
        )
        # Parent control (has source_citation)
        mock_parent_row = MagicMock(
            id="uuid-reg-1",
            control_id="SEC-042",
            title="Verschluesselung personenbezogener Daten",
            source_citation={"source": "DSGVO (EU) 2016/679", "article": "Art. 32"},
            parent_control_uuid=None,
            severity="high",
            category="encryption",
        )

        mock_qdrant_results = [
            {
                "score": 0.89,
                "payload": {
                    "control_uuid": "uuid-atomic-1",
                    "control_id": "SEC-042-A01",
                    "title": "Verschluesselung (atomar)",
                },
            },
            {
                "score": 0.65,  # Below threshold
                "payload": {
                    "control_uuid": "uuid-reg-2",
                    "control_id": "SEC-100",
                },
            },
        ]

        with patch("compliance.services.v1_enrichment.SessionLocal") as mock_session:
            db = MagicMock()
            mock_session.return_value.__enter__ = MagicMock(return_value=db)
            mock_session.return_value.__exit__ = MagicMock(return_value=False)

            # Route queries to correct mock data
            def side_effect_execute(query, params=None):
                result = MagicMock()
                query_str = str(query)
                result.fetchall.return_value = mock_v1
                if "COUNT" in query_str:
                    result.fetchone.return_value = mock_count
                elif "source_citation IS NOT NULL" in query_str:
                    # Parent lookup
                    result.fetchone.return_value = mock_parent_row
                elif "c.id = CAST" in query_str or "canonical_controls c" in query_str:
                    # Direct atomic control lookup
                    result.fetchone.return_value = mock_atomic_row
                else:
                    result.fetchone.return_value = mock_count
                return result

            db.execute.side_effect = side_effect_execute

            with patch("compliance.services.v1_enrichment.get_embedding") as mock_embed, \
                 patch("compliance.services.v1_enrichment.qdrant_search_cross_regulation") as mock_qdrant:
                mock_embed.return_value = [0.1] * 1024
                mock_qdrant.return_value = mock_qdrant_results

                result = await enrich_v1_matches(dry_run=False, batch_size=100, offset=0)

        assert result["dry_run"] is False
        assert result["processed"] == 1
        assert result["matches_inserted"] == 1
        assert len(result["sample_matches"]) == 1
        assert result["sample_matches"][0]["matched_control_id"] == "SEC-042"
        assert result["sample_matches"][0]["similarity_score"] == 0.89

    @pytest.mark.asyncio
    async def test_empty_batch_returns_done(self):
        mock_count = MagicMock(cnt=863)

        with patch("compliance.services.v1_enrichment.SessionLocal") as mock_session:
            db = MagicMock()
            mock_session.return_value.__enter__ = MagicMock(return_value=db)
            mock_session.return_value.__exit__ = MagicMock(return_value=False)
            db.execute.return_value.fetchall.return_value = []
            db.execute.return_value.fetchone.return_value = mock_count

            result = await enrich_v1_matches(dry_run=False, batch_size=100, offset=9999)

        assert result["processed"] == 0
        assert "alle v1 Controls verarbeitet" in result["message"]


class TestV1MatchesEndpoint:
    """Test the matches retrieval."""

    @pytest.mark.asyncio
    async def test_returns_matches(self):
        mock_rows = [
            MagicMock(
                matched_control_id="SEC-042",
                matched_title="Verschluesselung",
                matched_objective="Daten verschluesseln",
                matched_severity="high",
                matched_category="encryption",
                matched_source="DSGVO (EU) 2016/679",
                matched_article="Art. 32",
                matched_source_citation={"source": "DSGVO (EU) 2016/679"},
                similarity_score=0.89,
                match_rank=1,
                match_method="embedding",
            ),
        ]

        with patch("compliance.services.v1_enrichment.SessionLocal") as mock_session:
            db = MagicMock()
            mock_session.return_value.__enter__ = MagicMock(return_value=db)
            mock_session.return_value.__exit__ = MagicMock(return_value=False)
            db.execute.return_value.fetchall.return_value = mock_rows

            result = await get_v1_matches("uuid-v1-1")

        assert len(result) == 1
        assert result[0]["matched_control_id"] == "SEC-042"
        assert result[0]["similarity_score"] == 0.89
        assert result[0]["matched_source"] == "DSGVO (EU) 2016/679"

    @pytest.mark.asyncio
    async def test_empty_matches(self):
        with patch("compliance.services.v1_enrichment.SessionLocal") as mock_session:
            db = MagicMock()
            mock_session.return_value.__enter__ = MagicMock(return_value=db)
            mock_session.return_value.__exit__ = MagicMock(return_value=False)
            db.execute.return_value.fetchall.return_value = []

            result = await get_v1_matches("uuid-nonexistent")

        assert result == []


class TestEigenentwicklungDetection:
    """Verify the Eigenentwicklung detection query."""

    @pytest.mark.asyncio
    async def test_count_v1_controls(self):
        mock_count = MagicMock(cnt=863)

        with patch("compliance.services.v1_enrichment.SessionLocal") as mock_session:
            db = MagicMock()
            mock_session.return_value.__enter__ = MagicMock(return_value=db)
            mock_session.return_value.__exit__ = MagicMock(return_value=False)
            db.execute.return_value.fetchone.return_value = mock_count

            result = await count_v1_controls()

        assert result == 863
        # Verify the query includes all conditions
        call_args = db.execute.call_args[0][0]
        query_str = str(call_args)
        assert "generation_strategy = 'ungrouped'" in query_str
        assert "source_citation IS NULL" in query_str
        assert "parent_control_uuid IS NULL" in query_str