tests/uat/lib/harness-page-driver.ts

@@ -40,12 +40,45 @@ import { join, resolve as resolvePath } from 'node:path';
 
 import JSZip from 'jszip';
 import { EventType, IncrementalSource } from '@rrweb/types';
-import type { Page } from 'puppeteer';
+import type { Browser, Page } from 'puppeteer';
 
 import type { AssertionRecord, CheckRecord } from './assertions';
 import { assertArchiveShape, extractEntryToFile } from './zip';
 import type { UserEvent } from '../../../src/shared/types';
 
+/**
+ * Force-terminate the MV3 service worker via Puppeteer CDP. Required
+ * because Puppeteer's persistent CDP attach keeps SWs alive indefinitely;
+ * natural 30s idle eviction does NOT fire under test conditions per
+ * Chrome docs.
+ *
+ * Plan 04-04 — Wave 0 SPIKE persisting artifact + Wave 1 driveA33
+ * helper. The Puppeteer ≥ 22.1.0 `WebWorker.close()` surface is the
+ * canonical Chrome devrel pattern for simulating MV3 SW idle eviction
+ * under Puppeteer-driven UAT (CDP attach prevents natural eviction).
+ *
+ * References:
+ *   - https://developer.chrome.com/docs/extensions/how-to/test/test-serviceworker-termination-with-puppeteer
+ *   - https://developer.chrome.com/docs/extensions/develop/concepts/service-workers/lifecycle
+ *   - https://developer.chrome.com/blog/eyeos-journey-to-testing-mv3-service%20worker-suspension
+ *
+ * @param browser     - Puppeteer Browser handle from `launchHarnessBrowser`.
+ * @param extensionId - The runtime extension ID (from `handles.extensionId`).
+ */
+export async function stopServiceWorker(
+  browser: Browser,
+  extensionId: string,
+): Promise<void> {
+  const host = `chrome-extension://${extensionId}`;
+  const target = await browser.waitForTarget(
+    (t) => t.type() === 'service_worker' && t.url().startsWith(host),
+  );
+  const worker = await target.worker();
+  if (worker !== null) {
+    await worker.close();
+  }
+}
+
 /**
  * Extended assertion-record shape for A5/A12/A13 which return
  * host-side-required binary payloads:
@@ -1384,17 +1417,21 @@ const A27_HOST_POLL_TIMEOUT_MS = 8_000;
 const A27_HOST_POLL_INTERVAL_MS = 100;
 
 /**
- * Internal: pick the most-recently-modified .zip file in `downloadsDir`,
+ * Pick the most-recently-modified .zip file in `downloadsDir`, or null
- * or null if no .zip files exist. Used by driveA26 + driveA28 to chain
+ * if no .zip files exist. Used by driveA26 + driveA28 to chain off
- * off A25/A27 without re-dispatching SAVE. The "latest mtime wins"
+ * A25/A27 without re-dispatching SAVE. The "latest mtime wins" pattern
- * pattern works because the A25/A27 driver returns AFTER its zip lands
+ * works because the A25/A27 driver returns AFTER its zip lands
  * (await-pattern via the stable-size poll); no race with mid-write
  * partial zips at the read instant.
  *
+ * Plan 04-04 — exported so the Wave 0 spike script
+ * (`tests/uat/spike-a33-sw-persistence.ts`) can reuse the canonical
+ * mtime-sort archive-selection logic without duplicating it.
+ *
  * @param downloadsDir - Absolute path to the per-run downloads dir.
  * @returns Absolute path of the latest .zip, or null on empty dir.
  */
-function findLatestZip(downloadsDir: string): string | null {
+export function findLatestZip(downloadsDir: string): string | null {
   const candidates = readdirSync(downloadsDir).filter(isZipFilename);
   if (candidates.length === 0) {
     return null;

tests/uat/spike-a33-sw-persistence.ts

@@ -0,0 +1,201 @@
+// tests/uat/spike-a33-sw-persistence.ts — Plan 04-04 Wave 0 SPIKE.
+//
+// One-shot empirical investigation: does the offscreen document survive
+// a 5-min idle window followed by Puppeteer-driven force-termination
+// of the MV3 service worker? Per RESEARCH Q2 (sub-question b), the
+// canonical architecture analysis says yes (offscreen has its own
+// lifecycle, anchored by active MediaRecorder); RESEARCH confidence
+// is MEDIUM because the Chrome docs leave the offscreen-vs-SW interplay
+// implicit. This spike is the empirical hedge.
+//
+// Outcome decision tree:
+//   - videoSize > 100_000 bytes → SPIKE PASSED. Plan 04-04 Wave 1
+//     proceeds: A33 is a verification-only harness assertion that
+//     wraps the spike methodology. ROADMAP SC #1 satisfied by the
+//     CURRENT architecture; no persistence layer needed.
+//   - videoSize <= 100_000 bytes OR throw → SPIKE FAILED. Plan 04-04
+//     STOPS execution; SUMMARY documents the failure mode; the next
+//     planning event proposes IndexedDB persistence work (out of
+//     scope for this plan per the spike-first contract).
+//
+// Architectural reuse:
+//   - `setupFreshRecording` is module-internal inside
+//     `extension-page-harness.ts` (NOT exposed on `__mokoshHarness`).
+//     The exposed `assertA2` method is the canonical "prime fresh
+//     recording" entrypoint — it calls `ensureOffscreen + startRecording`
+//     and waits for `badge === 'REC'`. After `assertA2` returns,
+//     `MediaRecorder.state === 'recording'` and the offscreen is the
+//     "compelling reason" Chrome keeps it alive for. This matches the
+//     existing precedent: A3+A4 chain off A2's REC state.
+//   - SAVE_ARCHIVE dispatch uses `chrome.runtime.sendMessage` directly
+//     from the harness page realm (which is an extension-page realm
+//     with full chrome.* access). This matches the canonical pattern
+//     used by 9 existing assertA* methods (A5/A11/A12/A13/A26/A28/
+//     A29/A30/A31). The __mokoshHarness surface is assertA1..A31 +
+//     getManifestVersion only (no SAVE-dispatch helper) — Option B
+//     per Plan 04-04 REVISION iter-2.
+//
+// References:
+//   - Plan 04-04 PLAN.md <tasks> Task 1
+//   - .planning/phases/04-harden-clean-up-optional/04-RESEARCH.md Q2
+//   - https://developer.chrome.com/docs/extensions/how-to/test/test-serviceworker-termination-with-puppeteer
+//
+// Operation: tsx tests/uat/spike-a33-sw-persistence.ts (HEADLESS=1 for CI).
+// Expected wall-clock: ~6-7 minutes (5 min idle + 1-2 min orchestration).
+
+import { readFileSync } from 'node:fs';
+
+import JSZip from 'jszip';
+
+import { launchHarnessBrowser } from './lib/launch';
+import { findLatestZip, stopServiceWorker } from './lib/harness-page-driver';
+
+/** Wall-clock idle window. 5 min matches ROADMAP SC #1's "5+ min idle". */
+const SPIKE_IDLE_WAIT_MS = 5 * 60 * 1000;
+/** Post-`worker.close()` settle so the SW teardown finishes. */
+const SPIKE_NEW_SW_BOOT_MS = 500;
+/** SAVE_ARCHIVE round-trip timeout. */
+const SPIKE_SAVE_ARCHIVE_TIMEOUT_MS = 15_000;
+/** Post-SAVE settle so chrome.downloads finishes writing the zip. */
+const SPIKE_DOWNLOAD_SETTLE_MS = 5_000;
+/** Pass/fail floor: real archives are 1-3 MB; 100 KB is a sanity floor
+ *  above any "near-empty" failure mode (a near-empty zip with just
+ *  meta.json + screenshot is ~50 KB; anything above 100 KB means the
+ *  video buffer actually contained segments). */
+const SPIKE_VIDEO_SIZE_FLOOR_BYTES = 100_000;
+
+/**
+ * Spike entrypoint. Returns a Node exit code (0 = PASSED, 1 = FAILED).
+ */
+async function main(): Promise<number> {
+  process.stdout.write('\nMokosh Plan 04-04 Wave 0 SPIKE — SW state persistence empirical test\n');
+  process.stdout.write('='.repeat(72) + '\n');
+  process.stdout.write(
+    `SPIKE: idle window = ${SPIKE_IDLE_WAIT_MS}ms (${SPIKE_IDLE_WAIT_MS / 60_000} min); ` +
+      `pass floor = ${SPIKE_VIDEO_SIZE_FLOOR_BYTES} bytes\n\n`,
+  );
+
+  const t0 = Date.now();
+  const handles = await launchHarnessBrowser();
+  process.stdout.write(`SPIKE: launched browser; extensionId=${handles.extensionId}\n`);
+  process.stdout.write(`SPIKE: downloadsDir=${handles.downloadsDir}\n\n`);
+
+  let videoSize = 0;
+  let spikeError: string | null = null;
+
+  try {
+    // Step 1 — prime recording via assertA2. assertA2 is the canonical
+    // exposed "go to REC state" method on __mokoshHarness (read_first
+    // verified: __mokoshHarness has assertA1..A31 + getManifestVersion;
+    // assertA1 is read-only state inspection, NOT recording bootstrap;
+    // setupFreshRecording is module-internal, NOT exposed). assertA2
+    // does ensureOffscreen + startRecording + waitFor(badge==='REC').
+    process.stdout.write('SPIKE Step 1: prime recording via __mokoshHarness.assertA2\n');
+    const a2Result = await handles.harnessPage.evaluate(async () => {
+      const harness = (
+        window as unknown as {
+          __mokoshHarness: { assertA2: () => Promise<{ passed: boolean; error?: string }> };
+        }
+      ).__mokoshHarness;
+      return harness.assertA2();
+    });
+    process.stdout.write(
+      `SPIKE Step 1 result: assertA2.passed=${a2Result.passed}` +
+        (a2Result.error !== undefined ? `, error=${a2Result.error}` : '') +
+        '\n',
+    );
+    if (!a2Result.passed) {
+      throw new Error(`assertA2 priming failed: ${a2Result.error ?? '(no error)'}`);
+    }
+
+    // Step 2 — 5-min wall-clock idle. The whole point of the spike.
+    process.stdout.write(`\nSPIKE Step 2: waiting ${SPIKE_IDLE_WAIT_MS}ms for SW idle window...\n`);
+    process.stdout.write(`SPIKE Step 2: ETA ${new Date(Date.now() + SPIKE_IDLE_WAIT_MS).toISOString()}\n`);
+    await new Promise((res) => setTimeout(res, SPIKE_IDLE_WAIT_MS));
+    process.stdout.write('SPIKE Step 2 OK — 5-min idle elapsed\n');
+
+    // Step 3 — force-terminate the SW via Puppeteer CDP worker.close().
+    process.stdout.write('\nSPIKE Step 3: stopServiceWorker(browser, extensionId)\n');
+    await stopServiceWorker(handles.browser, handles.extensionId);
+    process.stdout.write('SPIKE Step 3 OK — SW terminated via worker.close()\n');
+
+    // Step 4 — brief settle for SW teardown.
+    await new Promise((res) => setTimeout(res, SPIKE_NEW_SW_BOOT_MS));
+
+    // Step 5 — dispatch SAVE_ARCHIVE via chrome.runtime.sendMessage from
+    // the harness page realm. The first message after worker.close()
+    // wakes the SW back up (event-driven respawn — canonical MV3 wakeup
+    // path). SAVE_ARCHIVE then triggers saveArchive() which calls
+    // getVideoBufferFromOffscreen() — if the offscreen survived, the
+    // segments array still has the accumulated 30s of webm chunks.
+    process.stdout.write('\nSPIKE Step 5: chrome.runtime.sendMessage({type: SAVE_ARCHIVE}) inline\n');
+    const saveResult = await handles.harnessPage.evaluate(
+      (timeoutMs: number) =>
+        new Promise<{ success: boolean; error?: string }>((resolve) => {
+          const timer = setTimeout(() => {
+            resolve({ success: false, error: `SAVE_ARCHIVE timed out after ${timeoutMs}ms` });
+          }, timeoutMs);
+          chrome.runtime.sendMessage({ type: 'SAVE_ARCHIVE' }, (response: unknown) => {
+            clearTimeout(timer);
+            if (chrome.runtime.lastError !== undefined) {
+              resolve({ success: false, error: String(chrome.runtime.lastError.message) });
+              return;
+            }
+            resolve(response as { success: boolean; error?: string });
+          });
+        }),
+      SPIKE_SAVE_ARCHIVE_TIMEOUT_MS,
+    );
+    process.stdout.write(`SPIKE Step 5 result: SAVE_ARCHIVE ack -> ${JSON.stringify(saveResult)}\n`);
+
+    // Step 6 — let chrome.downloads finish writing the file.
+    await new Promise((res) => setTimeout(res, SPIKE_DOWNLOAD_SETTLE_MS));
+
+    // Step 7 — find the produced zip + measure the video entry.
+    process.stdout.write('\nSPIKE Step 7: locate zip + inspect video/last_30sec.webm\n');
+    const zipPath = findLatestZip(handles.downloadsDir);
+    if (zipPath === null) {
+      throw new Error(`No zip found in downloadsDir=${handles.downloadsDir}`);
+    }
+    process.stdout.write(`SPIKE Step 7: zipPath=${zipPath}\n`);
+
+    const zip = await JSZip.loadAsync(readFileSync(zipPath));
+    const videoEntry = zip.file('video/last_30sec.webm');
+    videoSize = videoEntry !== null ? (await videoEntry.async('uint8array')).byteLength : 0;
+  } catch (err) {
+    spikeError = err instanceof Error ? err.message : String(err);
+    process.stderr.write(`\nSPIKE THREW: ${spikeError}\n`);
+  } finally {
+    try {
+      await handles.browser.close();
+    } catch (closeErr) {
+      process.stderr.write(`(non-fatal: browser close threw: ${String(closeErr)})\n`);
+    }
+  }
+
+  const tEnd = Date.now();
+  const elapsedSec = (tEnd - t0) / 1000;
+
+  process.stdout.write('\n' + '='.repeat(72) + '\n');
+  process.stdout.write(`SPIKE RESULT: videoSize=${videoSize} bytes (floor=${SPIKE_VIDEO_SIZE_FLOOR_BYTES}; elapsed=${elapsedSec.toFixed(1)}s)\n`);
+  if (spikeError !== null) {
+    process.stdout.write(`SPIKE OUTCOME: FAILED (spike threw: ${spikeError})\n`);
+    process.stdout.write('='.repeat(72) + '\n');
+    return 1;
+  }
+  if (videoSize > SPIKE_VIDEO_SIZE_FLOOR_BYTES) {
+    process.stdout.write('SPIKE OUTCOME: PASSED (offscreen SURVIVED the 5-min idle + SW kill)\n');
+    process.stdout.write('  ROADMAP SC #1 hypothesis empirically confirmed.\n');
+    process.stdout.write('  Plan 04-04 Wave 1 proceeds: A33 verification-only assertion.\n');
+    process.stdout.write('='.repeat(72) + '\n');
+    return 0;
+  }
+  process.stdout.write('SPIKE OUTCOME: FAILED (offscreen DIED — videoSize below floor)\n');
+  process.stdout.write('  Plan 04-04 Wave 1 BLOCKED: IndexedDB persistence work needed.\n');
+  process.stdout.write('  Route via plan-fix ceremony per saved-memory contract.\n');
+  process.stdout.write('='.repeat(72) + '\n');
+  return 1;
+}
+
+const code = await main();
+process.exit(code);