windows/couchcoding-logbert/tests/conftest.py

"""Recreation-Bench Windows pytest conftest — per-test app lifecycle + auto result collection.

Provides fixtures: app (per-test: launch → yield → kill), rb (per-test depth tracker),
results_dir, frame.

Each test's outcome is written into ``<module>_results.json`` incrementally (crash-safe).

Design notes:
- The ``app`` fixture launches a FRESH app instance for each test and kills it after.
- Cold mode (RB_APP_BINARY or RB_LAUNCH_SCRIPT set): per-test launch/kill with retries,
  PID-based window matching, image-name fallback kill.
- Warm mode (neither set): attach to already-running app, no kill on teardown.
- Results are flushed to disk after EVERY test (crash-safe).
"""
from __future__ import annotations

import json
import os
import shutil
import subprocess
import time

import pytest

try:
    import rb_testgen_kit as kit

    _tier_of = kit.tier_of
except Exception:
    def _tier_of(jk):
        return {"none": 0, "liveness": 1, "reach": 2, "transition": 2, "value": 3}.get(jk, 2)


# ── Results accumulator (flushed after every test for crash safety) ───────────
_results: list[dict] = []


def pytest_configure(config):
    config.addinivalue_line(
        "markers",
        "rb(jump_kind, expect, depth): declare behavioral metadata for scoring",
    )


def pytest_sessionstart(session):
    _results.clear()


def _extract_rb_marker(item):
    """(jump_kind, expect, declared_depth) from the @pytest.mark.rb(...) marker."""
    m = item.get_closest_marker("rb")
    if m is None:
        return "none", "", None
    jk = m.kwargs.get("jump_kind", m.args[0] if m.args else "none")
    expect = m.kwargs.get("expect", m.args[1] if len(m.args) > 1 else "")
    depth = m.kwargs.get("depth", None)
    return jk, expect, depth


# ── Depth tracker fixture ────────────────────────────────────────────────────
class _RBDepth:
    __slots__ = ("depth", "reached")

    def __init__(self):
        self.depth = 0
        self.reached = 0

    def enter(self, _surface: str = "") -> int:
        self.depth += 1
        self.reached = max(self.reached, self.depth)
        return self.depth

    def back(self) -> int:
        self.depth = max(0, self.depth - 1)
        return self.depth


@pytest.fixture
def rb(request):
    """Depth tracker: ``def test_x(app, rb): rb.enter("prefs")``. Fresh per test;
    pytest_runtest_makereport reads its ``reached`` into the record's ``depth``."""
    tracker = _RBDepth()
    request.node._rb = tracker
    return tracker


# ── App lifecycle helpers ────────────────────────────────────────────────────
def _results_dir() -> str:
    return os.environ.get("RB_RESULTS_DIR") or os.environ.get("RESULTS_DIR", ".")


_APP_STARTUP_WAIT = 3
_LAUNCH_RETRIES = 3


def _launch_app() -> int | None:
    """Launch the app with retries. Handles single-instance apps where the initial
    PID dies and the real process runs under a different PID (found by exe name).

    Returns PID or None.
    """
    launch_script = os.environ.get("RB_LAUNCH_SCRIPT", "")
    binary = os.environ.get("RB_APP_BINARY", "")
    exe_name = os.path.basename(binary) if binary else ""

    for attempt in range(_LAUNCH_RETRIES):
        pid = None

        if launch_script and os.path.exists(launch_script):
            try:
                ps_exe = shutil.which("pwsh") or "powershell"
                result = subprocess.run(
                    [ps_exe, "-ExecutionPolicy", "Bypass", "-File", launch_script],
                    capture_output=True, text=True, timeout=30,
                    encoding="utf-8", errors="replace",
                )
                for line in result.stdout.strip().splitlines():
                    line = line.strip()
                    if line.isdigit():
                        pid = int(line)
                        break
            except Exception:
                pass
        elif binary and os.path.exists(binary):
            try:
                proc = subprocess.Popen(
                    [binary], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL,
                )
                pid = proc.pid
            except Exception:
                pass

        if pid is None:
            if attempt < _LAUNCH_RETRIES - 1:
                time.sleep(3)
            continue

        # Wait for process to settle
        time.sleep(_APP_STARTUP_WAIT)

        # Verify the PID is still alive
        if _is_pid_alive(pid):
            return pid

        # Single-instance app: PID died, find real process by exe name
        if exe_name:
            real_pid = _find_pid_by_name(exe_name)
            if real_pid:
                return real_pid

        if attempt < _LAUNCH_RETRIES - 1:
            time.sleep(3)

    return None


def _is_pid_alive(pid: int) -> bool:
    """Check if a PID is still running."""
    try:
        result = subprocess.run(
            ["tasklist", "/FI", f"PID eq {pid}", "/NH", "/FO", "CSV"],
            capture_output=True, text=True, timeout=5,
        )
        return str(pid) in result.stdout
    except Exception:
        return False


def _find_pid_by_name(exe_name: str) -> int | None:
    """Find a running process by image name. Returns PID or None."""
    try:
        result = subprocess.run(
            ["tasklist", "/FI", f"IMAGENAME eq {exe_name}", "/NH", "/FO", "CSV"],
            capture_output=True, text=True, timeout=5,
        )
        for line in result.stdout.strip().splitlines():
            if exe_name.lower() in line.lower():
                parts = line.strip('"').split('","')
                if len(parts) >= 2 and parts[1].strip('"').isdigit():
                    return int(parts[1].strip('"'))
    except Exception:
        pass
    return None


def _kill_app(pid: int) -> None:
    """Graceful close first (prevents 'restore session?' on next launch),
    then force kill as fallback, then verify by image name."""
    binary = os.environ.get("RB_APP_BINARY", "")
    exe_name = os.path.basename(binary) if binary else ""

    # Step 1: Graceful close (WM_CLOSE) — gives app a chance to save state cleanly
    try:
        subprocess.run(
            ["taskkill", "/PID", str(pid)],
            capture_output=True, timeout=5,
        )
    except Exception:
        pass

    # Wait up to 3s for graceful exit
    for _ in range(6):
        if not _is_pid_alive(pid):
            break
        time.sleep(0.5)

    # Step 2: Force kill if still alive
    if _is_pid_alive(pid):
        try:
            subprocess.run(
                ["taskkill", "/F", "/T", "/PID", str(pid)],
                capture_output=True, timeout=10,
            )
        except Exception:
            pass
        time.sleep(1)

    # Step 3: Fallback kill by image name (catches single-instance app re-PID)
    if exe_name and _find_pid_by_name(exe_name):
        try:
            subprocess.run(
                ["taskkill", "/F", "/IM", exe_name],
                capture_output=True, timeout=10,
            )
        except Exception:
            pass
        time.sleep(1)


def _wait_for_window_by_pid(pid: int, timeout: float = 30.0):
    """Wait for a window belonging to the given PID (or its children) to appear."""
    try:
        from pywinauto import Desktop
        import psutil
    except ImportError:
        return None

    deadline = time.time() + timeout
    while time.time() < deadline:
        try:
            # Build PID tree (app may spawn child processes that own the window)
            pids = {pid}
            try:
                parent = psutil.Process(pid)
                for child in parent.children(recursive=True):
                    pids.add(child.pid)
            except (psutil.NoSuchProcess, psutil.AccessDenied):
                pass

            for w in Desktop(backend="uia").windows():
                try:
                    if w.process_id() in pids:
                        title = w.window_text() or ""
                        # Skip empty/system windows
                        if title and title not in ("", "Program Manager",
                                                   "Microsoft Text Input Application"):
                            return w
                except Exception:
                    continue
        except Exception:
            pass
        time.sleep(0.5)
    return None


# ── Stray system dialog dismissal (per-test, M5 fix) ────────────────────────
_STRAY_DIALOG_TITLES = {
    "Shutdown Event Tracker", "Windows Error Reporting",
    "Microsoft Visual C++ Runtime Library", "Windows Update",
    ".NET Framework", "Application Error", "Windows Security",
}


def _dismiss_stray_system_dialogs() -> None:
    """Close system-level dialogs that can steal focus or block interaction."""
    try:
        from pywinauto import Desktop
        from pywinauto.keyboard import send_keys
        for w in Desktop(backend="uia").windows():
            try:
                title = w.window_text() or ""
                if any(s in title for s in _STRAY_DIALOG_TITLES):
                    try:
                        w.close()
                    except Exception:
                        try:
                            w.set_focus()
                            send_keys("{ESC}")
                        except Exception:
                            pass
            except Exception:
                continue
    except Exception:
        pass


_post_launch_captured = False


def _get_module_from_argv() -> str:
    """Extract module name from sys.argv (pytest is invoked with the test file path)."""
    for arg in sys.argv:
        base = os.path.basename(arg)
        if base.startswith("test_uia_") and base.endswith(".py"):
            return base[len("test_uia_"):-len(".py")]
    return "unknown"


def _capture_post_launch(window) -> None:
    """Capture a desktop screenshot after the first successful app launch per module."""
    global _post_launch_captured
    if _post_launch_captured:
        return
    _post_launch_captured = True
    try:
        from PIL import ImageGrab
        rd = _results_dir()
        ss_dir = os.path.join(rd, "screenshots")
        os.makedirs(ss_dir, exist_ok=True)
        module = _get_module_from_argv()
        img = ImageGrab.grab()
        img.save(os.path.join(ss_dir, f"post_launch_{module}.png"))
    except Exception:
        pass


# ── App fixtures ─────────────────────────────────────────────────────────────
@pytest.fixture(scope="session")
def results_dir() -> str:
    return _results_dir()


@pytest.fixture
def app():
    """Per-test app lifecycle: launch fresh → yield window → kill.

    Cold mode: PID-based window matching (precise, no false positives).
    Warm mode: title-based matching (attaches to already-running app).
    """
    import rb_uia

    app_name = os.environ.get("RB_APP_NAME", "")
    binary = os.environ.get("RB_APP_BINARY", "")
    launch_script = os.environ.get("RB_LAUNCH_SCRIPT", "")

    # Dismiss system-level stray windows before each test (M5 fix)
    _dismiss_stray_system_dialogs()

    if binary or launch_script:
        # Cold mode: launch fresh app for this test
        pid = _launch_app()
        if pid is None:
            pytest.fail("app failed to launch (all retries exhausted)", pytrace=False)

        # Wait for window by PID (precise matching, no title ambiguity)
        w = _wait_for_window_by_pid(pid, timeout=30.0)
        if w is None:
            _kill_app(pid)
            pytest.fail(
                f"app launched (PID={pid}) but no window found within 30s",
                pytrace=False,
            )
        rb_uia.focus_app_window(w)
        # Capture post-launch screenshot (once per module for debugging)
        _capture_post_launch(w)
        yield w
        _kill_app(pid)
    else:
        # Warm mode: attach to already-running app (for run_live.ps1)
        w = rb_uia.get_app(app_name)
        if w is None:
            pytest.fail(
                f"app not found via UIA (RB_APP_NAME={app_name!r})",
                pytrace=False,
            )
        rb_uia.focus_app_window(w)
        yield w


@pytest.fixture
def frame(app):
    """Alias for the app's top-level window."""
    return app


# ── Hook: collect results per test + flush immediately (C3 fix) ──────────────
@pytest.hookimpl(hookwrapper=True)
def pytest_runtest_makereport(item, call):
    outcome = yield
    report = outcome.get_result()
    if call.when != "call" and not (call.when == "setup" and report.failed):
        return
    jk, expect, declared_depth = _extract_rb_marker(item)
    tracker = getattr(item, "_rb", None)
    measured = tracker.reached if tracker else 0
    depth = declared_depth if declared_depth is not None else measured

    passed = report.passed if call.when == "call" else False
    msg = ""
    if report.failed and report.longrepr:
        msg = str(report.longrepr)[-500:]

    rec = {
        "name": item.name,
        "passed": passed,
        "message": msg,
        "depth": depth,
        "jump_kind": jk,
        "tier": _tier_of(jk),
        "expect": expect,
    }
    _results.append(rec)
    # Flush immediately after each test (crash-safe: partial results survive)
    _flush_results(item)


def _flush_results(item=None):
    """Write current results to disk (incremental, crash-safe)."""
    rd = _results_dir()
    module = "unknown"
    if item:
        fspath = str(item.fspath.basename if hasattr(item.fspath, "basename")
                     else os.path.basename(str(item.fspath)))
        if fspath.startswith("test_uia_") and fspath.endswith(".py"):
            module = fspath[len("test_uia_"):-len(".py")]
    elif _results:
        module = _module_from_results()

    payload = {"module": module, "tests": list(_results)}
    try:
        os.makedirs(rd, exist_ok=True)
        with open(os.path.join(rd, f"{module}_results.json"), "w") as f:
            json.dump(payload, f, indent=2)
    except Exception:
        pass


def _module_from_results() -> str:
    return "unknown"


# ── Session finish: final write ──────────────────────────────────────────────
def pytest_sessionfinish(session, exitstatus):
    """Final write of results (covers the case where last test didn't trigger flush)."""
    rd = _results_dir()
    module = _module_from_session(session)
    payload = {"module": module, "tests": list(_results)}
    try:
        os.makedirs(rd, exist_ok=True)
        with open(os.path.join(rd, f"{module}_results.json"), "w") as f:
            json.dump(payload, f, indent=2)
    except Exception:
        pass


def _module_from_session(session) -> str:
    """Extract module name from the test file being run."""
    items = session.items
    if items:
        fspath = str(items[0].fspath.basename if hasattr(items[0].fspath, "basename")
                     else os.path.basename(str(items[0].fspath)))
        if fspath.startswith("test_uia_") and fspath.endswith(".py"):
            return fspath[len("test_uia_"):-len(".py")]
    return "unknown"


# ── Terminal summary: run self-audit ─────────────────────────────────────────
def pytest_terminal_summary(terminalreporter, exitstatus, config):
    rd = _results_dir()
    if not os.path.isdir(rd):
        return
    try:
        import rb_testgen_kit as k
        terminalreporter.write_sep("=", "RB Self-Audit")
        k.rb_audit(rd)
        k.rb_lint(rd)
        testgen_dir = os.path.dirname(rd) if os.path.basename(rd) == "baseline_results" else rd
        if os.path.isdir(testgen_dir):
            k.rb_lint_sources(testgen_dir)
            k.rb_audit_depth(testgen_dir, rd)
            k.rb_audit_tier(testgen_dir, rd)
    except Exception as e:
        terminalreporter.write_line(f"[rb_audit] skipped: {e}")