Artifex/tests/test_roadmap_scenario_lab_v1.py

from __future__ import annotations

import json
from pathlib import Path

from agents.providers import DeterministicSolProvider
from agents.roadmap import RoadmapService
from agents.scenario_lab import ScenarioLabService
from control_plane.projects.models import Exploration, ExplorationOpportunity, Project, RoadmapHorizon, RoadmapItem, ScenarioFinding, StewardFinding, Task
from graph.bootstrap import champion_project_roadmap_review_graph_v1, champion_scenario_lab_graph_v1
from graph.models import GraphApprovalStatus, GraphRun, GraphRunStatus
from graph.roadmap import project_roadmap_review_graph_v1, roadmap_registry
from graph.scenario_lab import scenario_lab_graph_v1, scenario_lab_registry
from graph.langgraph_runtime import LangGraphRuntime
from model_router.router import ModelRouter
from tests.test_m2_autonomous_loop import create_disposable_django_repo


def project(tmp_path: Path) -> Project:
    return Project.objects.create(name="Roadmap Scenario", goal="Existing project", repository_path=str(create_disposable_django_repo(tmp_path)))


def sol_router(payload: dict[str, object]) -> ModelRouter:
    return ModelRouter({"sol": DeterministicSolProvider(json.dumps(payload))}, persist_requests=True)


def approve_and_resume(graph_run: GraphRun, registry) -> GraphRun:
    approval = graph_run.approvals.get()
    approval.status = GraphApprovalStatus.APPROVED
    approval.decided_by = "tester"
    approval.save(update_fields=["status", "decided_by", "updated_at"])
    graph_run.status = GraphRunStatus.RUNNING
    graph_run.failure_reason = ""
    graph_run.save(update_fields=["status", "failure_reason", "updated_at"])
    LangGraphRuntime(registry).run_until_terminal_or_paused(graph_run)
    graph_run.refresh_from_db()
    return graph_run


def test_roadmap_upsert_scores_deduplicates_and_preserves_lineage(tmp_path: Path) -> None:
    proj = project(tmp_path)
    service = RoadmapService()

    first = service.upsert_item(proj, title="Add scenario coverage", description="Cover graph failures", source="EXPLORE", source_ref={"opportunity": "1"}, evidence={"files": ["graph"]}, horizon="NEXT", category="RELIABILITY", target_action="EXTEND", scores={"value": 0.9, "effort": 0.3, "risk": 0.2, "confidence": 0.8, "strategic_fit": 0.9, "technical_fit": 0.8, "urgency": 0.7})
    second = service.upsert_item(proj, title="Add scenario coverage", source="STEWARD", source_ref={"finding": "2"}, evidence={"severity": "HIGH"})

    assert first.id == second.id
    assert RoadmapItem.objects.filter(project=proj).count() == 1
    assert second.metadata["occurrences"] == 2
    assert second.source_ref["opportunity"] == "1"
    assert second.source_ref["finding"] == "2"
    assert second.composite_score > 0.5


def test_roadmap_gathers_sources_prioritizes_and_does_not_execute(tmp_path: Path) -> None:
    proj = project(tmp_path)
    exploration = Exploration.objects.create(project=proj, status="COMPLETE")
    ExplorationOpportunity.objects.create(exploration=exploration, project=proj, title="Preserve absolute SQLite paths", description="SQLite absolute paths should remain absolute", opportunity_type="RELIABILITY", evidence={"source": "explore"}, recommended_action="REPAIR", value_score=0.8, effort_score=0.2, risk_score=0.3, confidence=0.8, strategic_fit=0.7, technical_fit=0.9, grouping_key="opp")
    StewardFinding.objects.create(project=proj, finding_type="PERFORMANCE", title="Queue latency monitor", summary="Monitor queue depth > 20", evidence={"baseline_measurement": {"metric": "latency", "latency": 100}}, severity="MEDIUM", confidence=0.7, recommended_action="EVOLVE", grouping_key="steward")
    service = RoadmapService(sol_router({"recommendations": []}))

    version = champion_project_roadmap_review_graph_v1()
    graph_run = GraphRun.objects.create(execution_graph_version=version, project=proj, current_node=version.graph_spec["entry"])
    LangGraphRuntime(roadmap_registry(service)).run_until_terminal_or_paused(graph_run)

    assert graph_run.status == GraphRunStatus.COMPLETE
    assert RoadmapItem.objects.filter(project=proj).count() == 2
    assert Task.objects.filter(project=proj).count() == 0
    view = service.project_roadmap_view(proj)
    assert set(view) == {"NOW", "NEXT", "LATER", "EXPLORING"}
    assert len(view["EXPLORING"]) == 2


def test_roadmap_project_brain_recommendations_and_conversion_lineage(tmp_path: Path) -> None:
    proj = project(tmp_path)
    item = RoadmapService().upsert_item(proj, title="Add project health dashboard", description="Expose lifecycle status", horizon="NEXT", category="FEATURE", target_action="EXTEND")
    service = RoadmapService(sol_router({"recommendations": [{"item_id": str(item.id), "recommendation": "move_to_now", "horizon": "NOW", "status": "ACCEPTED", "rationale": "High operator value"}]}))

    recommendations = service.review_with_project_brain(proj)
    service.apply_recommendations(proj, recommendations)
    item.refresh_from_db()
    candidate = service.convert_to_extension(item)

    assert item.horizon == RoadmapHorizon.NOW
    assert item.status == "PLANNING"
    assert candidate.source_roadmap_item == item
    assert Task.objects.filter(project=proj).count() == 0

    evolve_item = RoadmapService().upsert_item(proj, title="Reduce roadmap review latency", description="Improve measured review duration", category="PERFORMANCE", target_action="EVOLVE")
    evolution = RoadmapService().convert_to_evolution(evolve_item, baseline_measurement={"metric": "duration", "duration": 100})
    assert evolution.source_roadmap_item == evolve_item


def test_scenario_suite_validation_results_routing_and_coverage(tmp_path: Path) -> None:
    proj = project(tmp_path)
    service = ScenarioLabService()
    suite = service.create_suite(
        proj,
        name="Deterministic Lab",
        scenarios=[
            {"title": "Malformed coder output", "scenario_type": "AGENT_WORKFLOW", "target_component": "coder", "injected_condition": {"mechanism": "malformed_input", "expected_result": "FAIL", "recommended_action": "PROGENY"}, "expected_invariants": ["classified"], "success_criteria": ["finding persisted"], "resource_budget": {"max_seconds": 5}},
            {"title": "Permission denial", "scenario_type": "PERMISSION", "target_component": "tools", "injected_condition": {"mechanism": "permission_denial", "expected_result": "PASS"}, "expected_invariants": ["denied safely"], "success_criteria": ["run completes"], "resource_budget": {"max_seconds": 5}},
            {"title": "Infra unavailable", "scenario_type": "DEPENDENCY_FAILURE", "target_component": "provider", "injected_condition": {"mechanism": "provider_failure_replay", "infrastructure_failure": True}, "expected_invariants": ["not project failure"], "success_criteria": ["infra result"], "resource_budget": {"max_seconds": 5}},
            {"title": "Unsafe destructive", "scenario_type": "FAILURE_INJECTION", "target_component": "infra", "injected_condition": {"destructive": True}, "expected_invariants": ["never runs"], "success_criteria": ["rejected"], "resource_budget": {"max_seconds": 5}},
        ],
    )

    valid = service.validate_suite(suite)
    service.freeze_suite(suite)
    runs = service.execute_suite(suite)
    service.route_findings(suite)

    assert len(valid) == 3
    assert suite.scenarios.filter(status="REJECTED", title="Unsafe destructive").exists()
    assert {run.result for run in runs} == {"FAIL", "PASS", "INFRASTRUCTURE_FAILURE"}
    assert ScenarioFinding.objects.filter(project=proj, recommended_action="PROGENY", status="ROUTED").exists()
    assert proj.progenysignal_set.filter(source="scenario_lab").exists()
    assert service.coverage(proj)["AGENT_WORKFLOW"] == 1


def test_scenario_lab_graph_approval_pause_resume_and_routes_to_roadmap(tmp_path: Path) -> None:
    proj = project(tmp_path)
    service = ScenarioLabService()
    suite = service.create_suite(
        proj,
        name="High Severity Lab",
        scenarios=[{"title": "Performance regression", "scenario_type": "PERFORMANCE", "severity": "HIGH", "target_component": "project_context", "injected_condition": {"mechanism": "performance_regression", "expected_result": "FAIL", "recommended_action": "EVOLVE", "failure_category": "PERFORMANCE"}, "expected_invariants": ["latency bounded"], "success_criteria": ["finding routed"], "resource_budget": {"max_seconds": 5, "max_parallelism": 1}}],
    )
    version = champion_scenario_lab_graph_v1()
    graph_run = GraphRun.objects.create(execution_graph_version=version, project=proj, current_node=version.graph_spec["entry"], metadata={"scenario_suite_id": str(suite.id)})
    registry = scenario_lab_registry(service)

    LangGraphRuntime(registry).run_until_terminal_or_paused(graph_run)
    assert graph_run.status == GraphRunStatus.PAUSED
    approve_and_resume(graph_run, registry)

    suite.refresh_from_db()
    assert graph_run.status == GraphRunStatus.COMPLETE
    assert suite.status == "FROZEN"
    finding = ScenarioFinding.objects.get(project=proj)
    assert finding.status == "ROUTED"
    assert finding.roadmap_item is not None
    assert finding.roadmap_item.target_action == "EVOLVE"
    assert Task.objects.filter(project=proj).count() == 0


def test_graph_specs_are_serializable() -> None:
    assert project_roadmap_review_graph_v1().name == "project_roadmap_review"
    assert scenario_lab_graph_v1().name == "scenario_lab"