From b6b4081d9f83381c1cba5df7df272028c63d5adc Mon Sep 17 00:00:00 2001
From: Jammy2211 <JNightingale2211@gmail.com>
Date: Sat, 11 Jul 2026 19:22:06 +0100
Subject: [PATCH] =?UTF-8?q?feat:=20workspace=5Ftestmode=5Ftiming=20leg=20?=
 =?UTF-8?q?=E2=80=94=20TEST=5FMODE=20script=20timing=20(#65)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Phase 4b of the hygiene perf mode. Times a curated set of workspace start_here
scripts run in integration test mode (PYAUTO_TEST_MODE=2 +
PYAUTO_WORKSPACE_SMALL_DATASETS=1) — the developer-loop mode named in the
original hygiene prompt — and tracks slow-script regressions. Distinct from
script_timing (autobuild release-mode run_all).

- heart/checks/workspace_testmode_timing.py: runs the curated scripts in a
  SUBPROCESS with the test-mode env (rule 4; runner injected for stdlib-only
  tests). Rolling per-script baseline (~/.pyauto-heart/timings-ws-testmode/),
  ratio classify (1.5/3.0), writes ~/.pyauto-heart/workspace_testmode_timing.json.
  OFF-TICK (running scripts costs seconds-minutes → daily cron / on demand).
- state.py + dashboard.py: advisory "Workspace test-mode timing" section
  (local-only family; anchored away from the import_time/unit_test_timing edits
  to avoid merge conflicts); NOT in the readiness gating set.
- tests/test_workspace_testmode_timing.py: 6 stdlib-only tests (injected runner).

hygiene perf already reads workspace_testmode_timing.json (PyAutoBrain#94), so
this lights up in perf with no conductor change. Full suite 257 pass.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Tw3EwV55k6VzxorYng3Kfn
---
 heart/checks/workspace_testmode_timing.py | 210 ++++++++++++++++++++++
 heart/dashboard.py                        |  25 +++
 heart/state.py                            |   1 +
 tests/test_workspace_testmode_timing.py   |  77 ++++++++
 4 files changed, 313 insertions(+)
 create mode 100644 heart/checks/workspace_testmode_timing.py
 create mode 100644 tests/test_workspace_testmode_timing.py

diff --git a/heart/checks/workspace_testmode_timing.py b/heart/checks/workspace_testmode_timing.py
new file mode 100644
index 0000000..b7b7b38
--- /dev/null
+++ b/heart/checks/workspace_testmode_timing.py
@@ -0,0 +1,210 @@
+"""heart/checks/workspace_testmode_timing.py — time curated workspace scripts run
+in integration TEST MODE, track a rolling baseline, flag regressions.
+
+Distinct from ``script_timing``, which reads PyAutoBuild ``run_all`` durations
+(the release/CI path). This leg runs a curated set of workspace ``start_here``
+scripts with ``PYAUTO_TEST_MODE=2`` + ``PYAUTO_WORKSPACE_SMALL_DATASETS=1`` — the
+developer-loop mode named in the original hygiene prompt — and times them.
+
+OFF-TICK by design: running scripts costs seconds–minutes, which does not fit
+the <30s watch-loop budget (``docs/internals.md`` rule 3). Run on a slower
+cadence (daily cron / on demand), NOT from ``heart/tick.sh``:
+
+    python -m heart.checks.workspace_testmode_timing
+
+Scripts run in a SUBPROCESS (rule 4 — Heart never imports the science stack; the
+runner is injected so the stdlib-only tests use fixtures). Advisory: surfaced on
+the board but never a release gate.
+
+State: ~/.pyauto-heart/timings-ws-testmode/<slug>.json — rolling window per script.
+Output: ~/.pyauto-heart/workspace_testmode_timing.json — regression summary.
+Classification mirrors script_timing (ratio = latest / median(prior window)).
+"""
+
+from __future__ import annotations
+
+import json
+import os
+import re
+import statistics
+import subprocess
+import sys
+import time
+from pathlib import Path
+from typing import Any, Callable
+
+import yaml
+
+HEART_STATE_DIR = Path(
+    os.environ.get("HEART_STATE_DIR")
+    or Path.home() / ".pyauto-heart"
+)
+HEART_WS_TIMINGS_DIR = HEART_STATE_DIR / "timings-ws-testmode"
+HEART_HOME = Path(__file__).resolve().parents[2]
+CONFIG_PATH = HEART_HOME / "config" / "repos.yaml"
+
+# Curated dev-loop scripts (the smoke-subset spirit — do NOT grow this to make
+# the signal feel stronger). (workspace_repo, workspace-relative script path).
+DEFAULT_SCRIPTS = [
+    ("autolens_workspace", "scripts/imaging/start_here.py"),
+    ("autogalaxy_workspace", "scripts/imaging/start_here.py"),
+    ("autofit_workspace", "scripts/overview/overview_1_the_basics.py"),
+]
+
+# A runner maps (workspace, rel_path) to the script's wall-clock in seconds, or
+# None when it could not be timed (missing / errored / timed out). Injected in tests.
+Runner = Callable[[str, str], "float | None"]
+
+
+def load_thresholds() -> tuple[float, float, int]:
+    yellow, red, window = 1.5, 3.0, 7
+    if CONFIG_PATH.is_file():
+        cfg = yaml.safe_load(CONFIG_PATH.read_text()) or {}
+        t = cfg.get("thresholds", {}).get("workspace_testmode_timing", {})
+        yellow = float(t.get("yellow_factor", yellow))
+        red = float(t.get("red_factor", red))
+        window = int(t.get("baseline_window", window))
+    return yellow, red, window
+
+
+def default_runner(root: Path, python: str, timeout: float) -> Runner:
+    """Time ``python <script>`` in the workspace, in TEST_MODE, in a subprocess."""
+    env = {
+        **os.environ,
+        "PYAUTO_TEST_MODE": "2",
+        "PYAUTO_WORKSPACE_SMALL_DATASETS": "1",
+        "NUMBA_CACHE_DIR": os.environ.get("NUMBA_CACHE_DIR", "/tmp/numba_cache"),
+        "MPLCONFIGDIR": os.environ.get("MPLCONFIGDIR", "/tmp/matplotlib"),
+        "PYTHONUNBUFFERED": "1",
+    }
+
+    def run_script(workspace: str, rel: str) -> float | None:
+        workspace_dir = root / workspace
+        script = workspace_dir / rel
+        if not script.is_file():
+            return None
+        start = time.perf_counter()
+        try:
+            proc = subprocess.run(
+                [python, str(script)], cwd=str(workspace_dir),
+                capture_output=True, timeout=timeout, env=env,
+            )
+        except (subprocess.TimeoutExpired, FileNotFoundError):
+            return None
+        if proc.returncode != 0:
+            return None
+        return time.perf_counter() - start
+
+    return run_script
+
+
+def _slug(workspace: str, rel: str) -> str:
+    safe = re.sub(r"[^A-Za-z0-9]+", "_", f"{workspace}__{rel}").strip("_")
+    return f"{safe}.json"
+
+
+def update_history(slug: str, duration: float, window: int) -> list[float]:
+    HEART_WS_TIMINGS_DIR.mkdir(parents=True, exist_ok=True)
+    history_path = HEART_WS_TIMINGS_DIR / slug
+    if history_path.is_file():
+        try:
+            history = json.loads(history_path.read_text())
+        except json.JSONDecodeError:
+            history = []
+    else:
+        history = []
+    history.append(duration)
+    history = history[-window:]
+    history_path.write_text(json.dumps(history))
+    return history
+
+
+def classify(ratio: float, yellow: float, red: float) -> str:
+    if ratio > red:
+        return "red"
+    if ratio > yellow:
+        return "yellow"
+    return "green"
+
+
+def run(runner: Runner | None = None, scripts: list[tuple[str, str]] | None = None) -> dict[str, Any]:
+    yellow_factor, red_factor, window = load_thresholds()
+    scripts = scripts if scripts is not None else DEFAULT_SCRIPTS
+    python = os.environ.get("HYGIENE_PYTHON", "python3")
+    timeout = float(os.environ.get("HEART_WS_SCRIPT_TIMEOUT", "300"))
+    root = Path(os.environ.get("PYAUTO_ROOT", str(Path.home() / "Code" / "PyAutoLabs")))
+    runner = runner or default_runner(root, python, timeout)
+
+    findings: dict[str, list[dict[str, Any]]] = {"red": [], "yellow": [], "green": []}
+    measured = 0
+    unavailable: list[str] = []
+    new_scripts = 0
+
+    for workspace, rel in scripts:
+        sid = f"{workspace}/{rel}"
+        duration = runner(workspace, rel)
+        if duration is None:
+            unavailable.append(sid)
+            continue
+        measured += 1
+        history = update_history(_slug(workspace, rel), duration, window)
+        if len(history) <= 1:
+            new_scripts += 1
+            continue
+        baseline = statistics.median(history[:-1])
+        if baseline <= 0:
+            continue
+        ratio = duration / baseline
+        findings[classify(ratio, yellow_factor, red_factor)].append({
+            "script": sid,
+            "latest_seconds": round(duration, 3),
+            "baseline_seconds": round(baseline, 3),
+            "ratio": round(ratio, 2),
+            "samples": len(history) - 1,
+        })
+
+    summary = {
+        "python": python,
+        "mode": "PYAUTO_TEST_MODE=2 PYAUTO_WORKSPACE_SMALL_DATASETS=1",
+        "scripts_measured": measured,
+        "scripts_unavailable": unavailable,
+        "new_scripts_no_baseline": new_scripts,
+        "red_count": len(findings["red"]),
+        "yellow_count": len(findings["yellow"]),
+        "green_count": len(findings["green"]),
+        "red": sorted(findings["red"], key=lambda x: -x["ratio"]),
+        "yellow": sorted(findings["yellow"], key=lambda x: -x["ratio"]),
+    }
+
+    sys.path.insert(0, str(HEART_HOME))
+    from heart import state
+
+    state.atomic_write_json(HEART_STATE_DIR / "workspace_testmode_timing.json", summary)
+    return summary
+
+
+def main(argv: list[str]) -> int:
+    summary = run()
+
+    from heart_color import c_ok, c_warn, c_fail, c_info, c_meta, glyph_ok, glyph_warn, glyph_fail
+
+    if summary["red_count"]:
+        glyph = glyph_fail()
+        label = c_fail(f"{summary['red_count']} red") + " " + c_warn(f"{summary['yellow_count']} yellow")
+    elif summary["yellow_count"]:
+        glyph = glyph_warn()
+        label = c_warn(f"{summary['yellow_count']} slower than baseline")
+    elif not summary["scripts_measured"]:
+        glyph = glyph_warn()
+        label = c_warn("no script runnable (set HYGIENE_PYTHON / PYAUTO_ROOT)")
+    else:
+        glyph = glyph_ok()
+        label = c_ok(f"{summary['green_count']} scripts within baseline")
+    extra = c_meta(f" ({summary['new_scripts_no_baseline']} new, no baseline)")
+    print(f"{glyph} {c_info('workspace_testmode_timing')} {label}{extra}")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
+    sys.exit(main(sys.argv))
diff --git a/heart/dashboard.py b/heart/dashboard.py
index b62ed18..0d202b6 100644
--- a/heart/dashboard.py
+++ b/heart/dashboard.py
@@ -59,6 +59,7 @@
     "import_time",
     "profiling_drift",
     "test_run",
+    "workspace_testmode_timing",
     "version_skew",
 )
 
@@ -423,6 +424,30 @@ def build_board(
                 Section("profiling_drift", "Profiling drift", st, summary, details)
             )
 
+    # Workspace test-mode script timing (advisory; off-tick) ------------------
+    if "workspace_testmode_timing" in unobserved:
+        sections.append(_unobs_section("workspace_testmode_timing", "Workspace test-mode timing"))
+    else:
+        wt = snapshot.get("workspace_testmode_timing") or {}
+        if wt:
+            r = _as_int(wt.get("red_count"))
+            y = _as_int(wt.get("yellow_count"))
+            g = _as_int(wt.get("green_count"))
+            if r:
+                st, summary = FAIL, f"{r} script regressions (>3× baseline), {y} slow (>1.5×)"
+            elif y:
+                st, summary = WARN, f"{y} scripts >1.5× baseline, {g} within"
+            elif not _as_int(wt.get("scripts_measured")):
+                st, summary = WARN, "no script runnable (set HYGIENE_PYTHON / PYAUTO_ROOT)"
+            else:
+                st, summary = OK, f"{g} scripts within baseline (TEST_MODE)"
+            details = [
+                f"✗ {e['script'].split('/')[-1]}  "
+                f"{e['latest_seconds']:.1f}s vs {e['baseline_seconds']:.1f}s ({e['ratio']}×)"
+                for e in (wt.get("red") or [])[:5]
+            ]
+            sections.append(Section("workspace_testmode_timing", "Workspace test-mode timing", st, summary, details))
+
     # Test run ---------------------------------------------------------------
     if "test_run" in unobserved:
         sections.append(_unobs_section("test_run", "Test run"))
diff --git a/heart/state.py b/heart/state.py
index d2650c1..14d5464 100644
--- a/heart/state.py
+++ b/heart/state.py
@@ -77,6 +77,7 @@ def aggregate() -> dict[str, Any]:
         "import_time": _read_json_or_default(HEART_STATE_DIR / "import_time.json", {}),
         "profiling_drift": _read_json_or_default(HEART_STATE_DIR / "profiling_drift.json", {}),
         "test_run": _read_json_or_default(HEART_STATE_DIR / "test_run.json", {}),
+        "workspace_testmode_timing": _read_json_or_default(HEART_STATE_DIR / "workspace_testmode_timing.json", {}),
         "version_skew": _read_json_or_default(HEART_STATE_DIR / "version_skew.json", {}),
         "manifest_drift": _read_json_or_default(HEART_STATE_DIR / "manifest_drift.json", {}),
         "verify_install": _read_json_or_default(HEART_STATE_DIR / "verify_install.json", {}),
diff --git a/tests/test_workspace_testmode_timing.py b/tests/test_workspace_testmode_timing.py
new file mode 100644
index 0000000..37fe5e0
--- /dev/null
+++ b/tests/test_workspace_testmode_timing.py
@@ -0,0 +1,77 @@
+"""tests/test_workspace_testmode_timing.py — TEST_MODE script-timing classifier.
+
+Stdlib-only (internals rule 4): the script runner is injected, so the suite
+never runs real workspace scripts or imports the science stack.
+"""
+
+from __future__ import annotations
+
+import importlib
+import json
+
+import pytest
+
+SCRIPTS = [("autolens_workspace", "scripts/imaging/modeling/start_here.py")]
+
+
+@pytest.fixture
+def tmp_state(tmp_path, monkeypatch):
+    monkeypatch.setenv("HEART_STATE_DIR", str(tmp_path))
+    import heart.state as state_mod
+    importlib.reload(state_mod)
+    import heart.checks.workspace_testmode_timing as wt
+    importlib.reload(wt)
+    wt.HEART_STATE_DIR = tmp_path
+    wt.HEART_WS_TIMINGS_DIR = tmp_path / "timings-ws-testmode"
+    wt.HEART_WS_TIMINGS_DIR.mkdir(parents=True, exist_ok=True)
+    return tmp_path, wt
+
+
+def _runner(value):
+    return lambda workspace, rel: value
+
+
+def test_first_observation_has_no_baseline(tmp_state):
+    _, wt = tmp_state
+    s = wt.run(runner=_runner(20.0), scripts=SCRIPTS)
+    assert s["new_scripts_no_baseline"] == 1
+    assert s["red_count"] == 0 and s["yellow_count"] == 0
+    assert s["scripts_measured"] == 1
+
+
+def test_within_baseline_green(tmp_state):
+    _, wt = tmp_state
+    wt.run(runner=_runner(20.0), scripts=SCRIPTS)
+    s = wt.run(runner=_runner(20.0), scripts=SCRIPTS)
+    assert s["green_count"] == 1 and s["red_count"] == 0 and s["yellow_count"] == 0
+
+
+def test_yellow_and_red_thresholds(tmp_state):
+    _, wt = tmp_state
+    wt.run(runner=_runner(20.0), scripts=SCRIPTS)
+    s = wt.run(runner=_runner(40.0), scripts=SCRIPTS)  # 2.0×
+    assert s["yellow_count"] == 1 and s["red_count"] == 0
+    s2 = wt.run(runner=_runner(200.0), scripts=SCRIPTS)  # >3× vs updated median
+    assert s2["red_count"] == 1
+
+
+def test_script_unavailable(tmp_state):
+    _, wt = tmp_state
+    s = wt.run(runner=_runner(None), scripts=SCRIPTS)
+    assert s["scripts_measured"] == 0
+    assert s["scripts_unavailable"] == ["autolens_workspace/scripts/imaging/modeling/start_here.py"]
+
+
+def test_rolling_window_caps_history(tmp_state):
+    _, wt = tmp_state
+    for d in [10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 16.0, 17.0, 18.0, 19.0]:
+        wt.run(runner=_runner(d), scripts=SCRIPTS)
+    files = list(wt.HEART_WS_TIMINGS_DIR.glob("*.json"))
+    assert len(files) == 1
+    assert len(json.loads(files[0].read_text())) == 7
+
+
+def test_summary_records_test_mode(tmp_state):
+    _, wt = tmp_state
+    s = wt.run(runner=_runner(20.0), scripts=SCRIPTS)
+    assert "PYAUTO_TEST_MODE" in s["mode"] and "SMALL_DATASETS" in s["mode"]