sanguo_vnpy_v2/scripts/smoke_phase3a.py

"""Phase 3a 端到端冒烟：异步 submit + WS 阶段 + JWT + tears（容器）。"""
import sys
import os
import asyncio
import tempfile
import shutil

# Set up paths for both local and container environments
if os.path.exists("/app"):
    sys.path.insert(0, "/app")
    sys.path.insert(0, "/app/vnpy_v4.4.0")
    print("=== Using container paths ===")
else:
    # Local development environment
    repo_root = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
    sys.path.insert(0, repo_root)
    vnpy_path = os.path.join(repo_root, "vnpy_v4.4.0")
    if vnpy_path not in sys.path:
        sys.path.insert(0, vnpy_path)
    print(f"=== Using local paths: {repo_root} ===")


def test_sys_path():
    """Test sys.path configuration"""
    print("=== SYS.PATH CONFIG ===")
    for i, p in enumerate(sys.path[:3]):
        print(f"  path[{i}]: {p}")
    print("=== PASS: sys.path configured ===")
    print()


def test_jwt_login():
    """Test JWT login endpoint"""
    print("=== JWT LOGIN TEST ===")
    try:
        from fastapi.testclient import TestClient
        from sanguo_api.app import create_app
        from sanguo_api.auth import hash_password

        # Create temporary directory for test
        tmpdir = tempfile.mkdtemp()
        db_path = os.path.join(tmpdir, "test.db")
        file_dir = tmpdir

        try:
            # Generate real password hash using bcrypt directly
            test_password = "password123"
            password_hash = hash_password(test_password)

            # Create app with auth config
            app = create_app(
                db_path=db_path,
                file_dir=file_dir,
                auth_config={
                    "username": "admin",
                    "password_hash": password_hash,
                    "jwt_secret": "test-secret",
                    "expire_minutes": 60
                },
                max_workers=1
            )

            # Test login with TestClient - REAL password verification
            client = TestClient(app)
            response = client.post("/api/v1/auth/login", json={
                "username": "admin",
                "password": test_password
            })

            assert response.status_code == 200, f"Expected 200, got {response.status_code}"
            data = response.json()
            assert "token" in data, "Token not in response"
            print(f"  Login successful, token: {data['token'][:20]}...")
            print("=== PASS: JWT login ===")
            return data['token']
        finally:
            shutil.rmtree(tmpdir, ignore_errors=True)
    except Exception as e:
        print(f"  === FAIL: JWT login test failed: {type(e).__name__}: {e} ===")
        import traceback
        traceback.print_exc()
        print("=== FAIL: JWT login ===")
        return None


def test_protected_route_auth():
    """Test protected route authentication"""
    print("=== PROTECTED ROUTE AUTH TEST ===")
    try:
        from fastapi.testclient import TestClient
        from sanguo_api.app import create_app
        from sanguo_api.auth import hash_password

        tmpdir = tempfile.mkdtemp()
        db_path = os.path.join(tmpdir, "test.db")
        file_dir = tmpdir

        try:
            # Generate real password hash using bcrypt directly
            test_password = "password123"
            password_hash = hash_password(test_password)

            app = create_app(
                db_path=db_path,
                file_dir=file_dir,
                auth_config={
                    "username": "admin",
                    "password_hash": password_hash,
                    "jwt_secret": "test-secret",
                    "expire_minutes": 60
                },
                max_workers=1
            )

            client = TestClient(app)

            # Test without token - should get 401
            response = client.get("/api/v1/task/foo")
            assert response.status_code == 401, f"Expected 401 without token, got {response.status_code}"
            print("  No token: 401 Unauthorized ✓")

            # Get valid token using REAL password verification
            login_response = client.post("/api/v1/auth/login", json={
                "username": "admin",
                "password": test_password
            })
            token = login_response.json()["token"]

            # Test with token but non-existent task - should get 404
            response = client.get(
                "/api/v1/task/foo",
                headers={"Authorization": f"Bearer {token}"}
            )
            assert response.status_code == 404, f"Expected 404 with valid token, got {response.status_code}"
            print("  Valid token, non-existent task: 404 Not Found ✓")
            print("=== PASS: protected route auth ===")
        finally:
            shutil.rmtree(tmpdir, ignore_errors=True)
    except Exception as e:
        print(f"  === FAIL: Protected route auth test failed: {type(e).__name__}: {e} ===")
        import traceback
        traceback.print_exc()
        print("=== FAIL: protected route auth ===")


async def test_orchestrator_async():
    """Test orchestrator async submission"""
    print("=== ORCHESTRATOR ASYNC TEST ===")
    from sanguo_orchestrator.runner import Orchestrator

    tmpdir = tempfile.mkdtemp()
    db_path = os.path.join(tmpdir, "test.db")

    try:
        orch = Orchestrator(db_path=db_path, file_dir=tmpdir, max_workers=1)

        # Check executor type
        from concurrent.futures import ProcessPoolExecutor
        executor_type = type(orch.pool.executor).__name__
        print(f"  Orchestrator pool executor: {executor_type}")
        assert executor_type == "ProcessPoolExecutor", f"Expected ProcessPoolExecutor, got {executor_type}"

        # Submit a tiny factor task
        task_id = await orch.submit_factor(
            symbols=["600000.SSE"],
            factor_names=["ma5"],
            start="2024-01-01",
            end="2024-06-30",
            cfg=None,  # Use default config for smoke test
            output_dir=tmpdir
        )

        print(f"  Task submitted: {task_id}")
        assert task_id.startswith("factor_"), f"Expected task_id to start with 'factor_', got {task_id}"

        # Check task status
        status = orch.get_status(task_id)
        print(f"  Task status: {status}")

        print("=== PASS: orchestrator async ===")
        return task_id
    finally:
        shutil.rmtree(tmpdir, ignore_errors=True)


async def test_ws_stage_wiring():
    """Test WebSocket stage callback wiring"""
    print("=== WS STAGE WIRING TEST ===")
    from sanguo_api.app import create_app
    from sanguo_api.ws import manager
    from sanguo_api.routes import get_orchestrator
    from sanguo_api.auth import hash_password

    tmpdir = tempfile.mkdtemp()
    db_path = os.path.join(tmpdir, "test.db")

    try:
        # Generate real password hash using bcrypt directly
        test_password = "password123"
        password_hash = hash_password(test_password)

        # Create app to trigger wiring
        app = create_app(
            db_path=db_path,
            file_dir=tmpdir,
            auth_config={
                "username": "admin",
                "password_hash": password_hash,
                "jwt_secret": "test-secret",
                "expire_minutes": 60
            },
            max_workers=1
        )

        # Check orchestrator has on_stage callback set
        orch = get_orchestrator()
        assert orch._on_stage is not None, "on_stage callback not set"

        print("  WS manager exists ✓")
        print("  on_stage callback wired ✓")
        print("=== PASS: WS stage wiring ===")
    finally:
        shutil.rmtree(tmpdir, ignore_errors=True)


async def test_real_factor_tears_pipeline():
    """Test real factor tears pipeline with multiprocessing guard"""
    print("=== REAL FACTOR TEARS PIPELINE TEST ===")

    # Check if we're in container with full dependencies
    try:
        import polars
        import alphalens
        print("  Container environment detected (polars + alphalens available)")
    except ImportError as e:
        print(f"  === SKIP: polars/alphalens not available ({e}) ===")
        return

    tmpdir = tempfile.mkdtemp()
    output_dir = os.path.join(tmpdir, "factor_output")

    try:
        from sanguo_factor.analyzer import run_factor_analysis
        from sanguo_data.config import load_config

        # Load config for database access
        if os.path.exists("/app/config/data_platform.yaml"):
            cfg = load_config("/app/config/data_platform.yaml")
            print("  Loaded database config")
        else:
            print("  === SKIP: database config not found ===")
            return

        # ≥2 symbols: alphalens IC is cross-sectional (1 symbol → empty bins → fails)
        symbols = ["600000", "000001", "300750"]  # BARE symbols (no .EXCHANGE suffix) for DB lookup
        factor_names = ["ma5"]   # Simple factor
        start = "2024-01-01"
        end = "2024-06-30"     # Use longer range for reliable IC (Phase 1 confirmed 541 bars)

        print(f"  Running factor analysis: {symbols}, {factor_names}, {start} to {end}")
        print("  This will test the multiprocessing pipeline...")

        # Run the analysis
        result = run_factor_analysis(
            symbols=symbols,
            factor_names=factor_names,
            start=start,
            end=end,
            cfg=cfg,
            output_dir=output_dir
        )

        print(f"  Analysis complete: {len(result.factor_names)} factors processed")
        print(f"  IC summary: {result.ic_summary}")
        print(f"  Report paths: {result.report_paths}")

        # Verify we got a result
        assert result is not None, "Result is None"
        assert len(result.factor_names) > 0, "No factors processed"

        # CRITICAL: Verify IC is non-empty (Root cause C fix)
        assert result.ic_summary, "ic_summary empty - no IC computed"
        assert "ma5" in result.ic_summary, "ma5 not in ic_summary"
        assert result.ic_summary["ma5"].get("ic"), "no IC data for ma5"
        print(f"  ✓ IC computed: {result.ic_summary['ma5']['ic']}")

        # Check if any reports were generated
        if result.report_paths:
            print(f"  Tears report generated: {result.report_paths}")
            for factor_name, path in result.report_paths.items():
                if os.path.exists(path.replace('.html', '.png')):  # Check for actual PNG file
                    print(f"    ✓ {factor_name}: {path}")
                else:
                    print(f"    ⚠ {factor_name}: {path} (file not found)")

        print("=== PASS: real factor tears pipeline ===")
    except Exception as e:
        print(f"  === FAIL: real tears pipeline failed with {type(e).__name__}: {e} ===")
        import traceback
        traceback.print_exc()
        raise
    finally:
        shutil.rmtree(tmpdir, ignore_errors=True)


async def main():
    """Run all Phase 3a smoke tests"""
    print("\n")
    print("=" * 60)
    print("PHASE 3A SMOKE TEST SUITE")
    print("=" * 60)
    print()

    # Test 1: sys.path configuration
    test_sys_path()

    # Test 2: JWT login
    test_jwt_login()

    # Test 3: Protected route authentication
    test_protected_route_auth()

    # Test 4: Orchestrator async submission
    await test_orchestrator_async()

    # Test 5: WebSocket stage wiring
    await test_ws_stage_wiring()

    # Test 6: Real factor tears pipeline (container only)
    await test_real_factor_tears_pipeline()

    print()
    print("=" * 60)
    print("PHASE 3A SMOKE TEST SUITE: ALL TESTS PASSED")
    print("=" * 60)
    print()


if __name__ == "__main__":
    # CRITICAL: multiprocessing guard for vnpy.alpha
    # This is required because vnpy.alpha's AlphaDataset.prepare_data() spawns
    # a multiprocessing.Pool, which fails without this guard on spawn platforms
    print("=== Running with multiprocessing guard ===")
    asyncio.run(main())