chore(01-11): wave-0 — install puppeteer + tsx, add vite.test.config + Tier-1 hook-leak grep gate

Task 1 of Plan 01-11 (Puppeteer UAT harness). - npm install --save-dev puppeteer@^25.0.2 tsx@^4 @types/node resolved: puppeteer@25.x, tsx@4.22.1, @types/node@25.8.0 pulls ~150MB Chromium binary at install time (T-1-11-03 — accepted, package-lock pins resolved hashes via @puppeteer/browsers). - package.json scripts: add build:test + test:uat (per RESEARCH §10 two-bundle orchestration); existing dev/build/preview/test untouched. - vite.test.config.ts: extends ./vite.config.ts via mergeConfig with mode:'test' + build.outDir:'dist-test' + emptyOutDir:true. Verified npm run build:test produces dist-test/ in 7.93s; npm run build keeps producing dist/ in 7.67s (no clobber). - tsconfig.json `include: ["src"]` already covers src/test-hooks/**/* via wildcard — no edit needed. - tests/background/no-test-hooks-in-prod-bundle.test.ts: Tier-1 gate mirroring sw-bundle-import.test.ts's execFile pattern. Greps the BUILT dist/ tree for 5 forbidden hook surfaces (one `it` per surface for granular failure isolation): __mokoshTest, simulateUserStop, getSegmentCount, setCurrentStream, setSegmentCountGetter. All 5 surfaces absent today (RED-then-GREEN polarity inverted — the gate is GREEN now and MUST stay GREEN after Task 2 lands the hooks). SKIP_BUILD=1 escape hatch for developer iteration. - .gitignore: add dist-test/ (no point versioning generated test bundle). Verification: - npx tsc --noEmit: exit 0 - npm run build: exit 0; dist/ populated (375.37 kB SW chunk) - npm run build:test: exit 0; dist-test/ populated (identical chunk sizes — the gated dynamic imports do not land until Task 2; this commit only proves the two-bundle plumbing) - SKIP_BUILD=1 npx vitest run tests/background/no-test-hooks-in-prod-bundle.test.ts: 6/6 GREEN (1 build-sanity + 5 forbidden-surface) - SKIP_BUILD=1 npx vitest run (full suite): 89/89 GREEN (83 baseline + 6 new Tier-1 surfaces = 89) Working-tree cleanup: a stale 5.4 MB tests/fixtures/last_30sec.webm (unrelated operator smoke regen present at session spawn) was stashed before running the baseline — it caused the webm-playback test to time out at 5s. After stashing back to HEAD's 1.9 MB fixture, baseline passes cleanly. Not committing the fixture restoration here (pre-existing working-tree state, not part of Task 1). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-17 21:42:46 +02:00
parent 2669ce38e7
commit 96fa8e8e11
5 changed files with 1599 additions and 1 deletions
--- a/tests/background/no-test-hooks-in-prod-bundle.test.ts
+++ b/tests/background/no-test-hooks-in-prod-bundle.test.ts
@@ -0,0 +1,255 @@
+// tests/background/no-test-hooks-in-prod-bundle.test.ts
+//
+// Tier-1 hook-leak gate (Plan 01-11 Task 1) — sibling of
+// `sw-bundle-import.test.ts`. Both gates inspect the BUILT `dist/`
+// artifact for an invariant the SOURCE alone cannot prove.
+//
+// What this gate enforces — the security-critical invariant T-1-11-01:
+//
+//   Plan 01-11 introduces test-only "hook" surfaces under `src/test-hooks/`
+//   that expose internal SW + offscreen state (captured chrome.* handler
+//   refs, MediaStream getter, simulated user-stopped-sharing trigger) to
+//   the Puppeteer harness via a global named `__mokoshTest`. The hooks
+//   ship in the TEST bundle (`dist-test/`) and MUST NOT ship in the
+//   PRODUCTION bundle (`dist/`) — leaking them would expose Bug B's
+//   `simulateUserStop` path + the captured `onStartup` handler ref to
+//   any page that can `eval` against the extension's SW.
+//
+//   The leak is prevented by a Vite mode gate: each hook import in
+//   src/background/index.ts + src/offscreen/recorder.ts is wrapped in
+//   `if (import.meta.env.MODE === 'test') { await import('../test-hooks/...'); }`.
+//   Vite statically replaces `import.meta.env.MODE` at build time
+//   (production mode → `'production'`); the `'production' === 'test'`
+//   comparison is a static dead branch and Rollup tree-shakes the
+//   `await import` away entirely. That tree-shake is what THIS GATE
+//   verifies — by greping the built artifact tree for the hook surface
+//   strings and asserting they are absent.
+//
+// Why a unit-level gate IN ADDITION TO the harness's assertion 0:
+//   The harness's assertion 0 runs only when the harness runs (`npm run
+//   test:uat`), which requires a Chrome download + ~90s wall clock. The
+//   unit gate runs as part of the regular `npm test` pass — every
+//   developer's pre-push hook + every CI vitest job catches the leak in
+//   <15s. Belt + suspenders per Plan 01-11 RESEARCH §6 + the orchestrator-
+//   loaded `feedback-pre-checkpoint-bundle-gates.md` memory: any future
+//   plan executor whose work surfaces a SW build MUST keep this gate
+//   GREEN before any operator-empirical checkpoint.
+//
+// Polarity note: the gate is GREEN today (no hooks land until Plan 01-11
+// Task 2) AND must STAY GREEN after Task 2 lands them. The test is
+// committed BEFORE the hooks ship so the invariant is asserted from day
+// one — eliminating any window-of-vulnerability where the production
+// bundle could carry leaked hooks unnoticed.
+//
+// Surface inventory enforced (each MUST be absent from any file under
+// dist/ once builds + future plans add to the hook tree):
+//   - `__mokoshTest`              — the global surface name itself
+//   - `simulateUserStop`          — Bug B simulate function (Plan 01-11 §3)
+//   - `getSegmentCount`           — Plan 01-11 Task 7 segments-count getter
+//   - `setCurrentStream`          — Plan 01-11 Task 2 offscreen wire
+//   - `setSegmentCountGetter`     — Plan 01-11 Task 7 offscreen wire
+//
+// Implementation mirrors `sw-bundle-import.test.ts`'s execFile pattern:
+//   - Spawn `npm run build` via execFile so the build is reproducible
+//     and the gate runs against a known-clean artifact.
+//   - Skip the build if `process.env.SKIP_BUILD === '1'` — developer
+//     escape hatch when iterating on the gate itself.
+//   - Recursively walk `dist/` reading files synchronously (the tree is
+//     small; ~10 chunks; ~500 KB total).
+//   - For each forbidden string, count total occurrences and report the
+//     offending file paths on failure.
+//
+// References:
+//   - Vite mode + `import.meta.env.MODE`:
+//     https://vite.dev/guide/env-and-mode.html
+//   - Rollup tree-shaking + dead-branch elimination:
+//     https://rollupjs.org/configuration-options/#treeshake
+//   - Node `child_process.execFile`:
+//     https://nodejs.org/api/child_process.html#child_processexecfilefile-args-options-callback
+//   - Node `fs.readdirSync` + `withFileTypes`:
+//     https://nodejs.org/api/fs.html#fsreaddirsyncpath-options
+
+import { execFile } from 'node:child_process';
+import { existsSync, readFileSync, readdirSync, statSync } from 'node:fs';
+import { resolve as resolvePath } from 'node:path';
+import { promisify } from 'node:util';
+
+import { describe, expect, it } from 'vitest';
+
+const execFileAsync = promisify(execFile);
+
+/**
+ * Surface strings the gate forbids in any file under `dist/`. Order is
+ * preserved in failure diagnostics so the report is stable across runs.
+ * Each entry's rationale lives in the file header above.
+ */
+const FORBIDDEN_HOOK_STRINGS: ReadonlyArray<string> = [
+  '__mokoshTest',
+  'simulateUserStop',
+  'getSegmentCount',
+  'setCurrentStream',
+  'setSegmentCountGetter',
+];
+
+/** How long the build child has to finish (`npm run build` is ~10s).
+ *  Generous cap; if it blows past this something else is wrong. */
+const BUILD_TIMEOUT_MS = 60_000;
+
+/** Absolute path to the production output directory. */
+const DIST_DIR = resolvePath(process.cwd(), 'dist');
+
+/**
+ * One match in one file. Held in a flat array per forbidden string so
+ * the failure message can enumerate every (file, count) pair.
+ */
+interface ForbiddenMatch {
+  readonly filePath: string;
+  readonly count: number;
+}
+
+/**
+ * Recursively collect every regular file under `root`. Returns absolute
+ * paths. Skips symlinks defensively (none expected in the Vite output
+ * tree, but cheap to guard against).
+ *
+ * @param root - Absolute directory path to walk.
+ * @returns Sorted list of absolute file paths under `root`.
+ */
+function listAllFilesRecursive(root: string): ReadonlyArray<string> {
+  const accumulator: string[] = [];
+  const stack: string[] = [root];
+  while (stack.length > 0) {
+    const dir = stack.pop()!;
+    const entries = readdirSync(dir, { withFileTypes: true });
+    for (const entry of entries) {
+      const fullPath = resolvePath(dir, entry.name);
+      if (entry.isSymbolicLink()) {
+        continue;
+      }
+      if (entry.isDirectory()) {
+        stack.push(fullPath);
+      } else if (entry.isFile()) {
+        accumulator.push(fullPath);
+      }
+    }
+  }
+  return accumulator.sort();
+}
+
+/**
+ * Count occurrences of `needle` inside the given file's text content.
+ * Returns 0 when the file is binary-ish (no occurrences of a likely-text
+ * sentinel character class). Vite emits JS/CSS/HTML/JSON — all UTF-8 —
+ * plus copies of PNG icons. We skip files whose extensions clearly mark
+ * them as binary so readFileSync('utf8') does not return mojibake that
+ * could accidentally match `needle`.
+ *
+ * @param filePath - Absolute file path to scan.
+ * @param needle   - Literal substring to count.
+ * @returns Total occurrences of `needle` in the file's text.
+ */
+function countOccurrencesInFile(filePath: string, needle: string): number {
+  const binaryExtensions = new Set(['.png', '.jpg', '.jpeg', '.gif', '.ico', '.webp', '.woff', '.woff2', '.ttf', '.otf']);
+  const dotIdx = filePath.lastIndexOf('.');
+  const ext = dotIdx >= 0 ? filePath.substring(dotIdx).toLowerCase() : '';
+  if (binaryExtensions.has(ext)) {
+    return 0;
+  }
+  const stat = statSync(filePath);
+  if (stat.size === 0) {
+    return 0;
+  }
+  const text = readFileSync(filePath, 'utf8');
+  let count = 0;
+  let from = 0;
+  for (;;) {
+    const idx = text.indexOf(needle, from);
+    if (idx < 0) {
+      break;
+    }
+    count += 1;
+    from = idx + needle.length;
+  }
+  return count;
+}
+
+/**
+ * Walk `dist/` and find every file containing `needle`. Returns an
+ * array of (file, count) pairs sorted by file path. Empty when the
+ * needle is absent — that is the GREEN-gate condition.
+ *
+ * @param needle - Literal substring to grep for.
+ * @returns List of matches; empty array on absence.
+ */
+function findMatchesInDist(needle: string): ReadonlyArray<ForbiddenMatch> {
+  const files = listAllFilesRecursive(DIST_DIR);
+  const matches: ForbiddenMatch[] = [];
+  for (const filePath of files) {
+    const count = countOccurrencesInFile(filePath, needle);
+    if (count > 0) {
+      matches.push({ filePath, count });
+    }
+  }
+  return matches;
+}
+
+/**
+ * Spawn `npm run build` in a child process; reject on non-zero exit OR
+ * on timeout. Inherits parent env (we want the same `NODE_OPTIONS` etc.
+ * the developer set), but suppresses Node experimental warnings to
+ * keep vitest's failure output readable.
+ *
+ * @returns void on success; throws with build stderr captured on failure.
+ */
+async function runProductionBuild(): Promise<void> {
+  await execFileAsync('npm', ['run', 'build'], {
+    timeout: BUILD_TIMEOUT_MS,
+    maxBuffer: 16 * 1024 * 1024,
+    env: { ...process.env, NODE_NO_WARNINGS: '1' },
+  });
+}
+
+describe('production bundle has no test-hook leaks (Tier-1 gate — T-1-11-01)', () => {
+  it('npm run build completes and dist/ exists with at least one chunk', async () => {
+    if (process.env.SKIP_BUILD !== '1') {
+      await runProductionBuild();
+    }
+    expect(
+      existsSync(DIST_DIR),
+      `dist/ missing at ${DIST_DIR}. Either npm run build failed or SKIP_BUILD=1 was set ` +
+        `without a pre-existing build. The hook-leak gate cannot run without a built artifact.`,
+    ).toBe(true);
+    const files = listAllFilesRecursive(DIST_DIR);
+    expect(
+      files.length,
+      `dist/ is empty after npm run build — the build produced no output, which is a different ` +
+        `regression class than a hook leak. Investigate before proceeding to the hook-leak assertion.`,
+    ).toBeGreaterThan(0);
+  });
+
+  for (const needle of FORBIDDEN_HOOK_STRINGS) {
+    it(`production bundle does not contain '${needle}' (T-1-11-01 surface)`, () => {
+      // If the build did not run in the previous test (SKIP_BUILD=1) AND
+      // dist/ is missing, surface a clear diagnostic instead of letting
+      // the recursive walk throw an obscure ENOENT.
+      if (!existsSync(DIST_DIR)) {
+        throw new Error(
+          `dist/ missing — run \`npm run build\` first (SKIP_BUILD=1 is set but no prior build artifact exists).`,
+        );
+      }
+      const matches = findMatchesInDist(needle);
+      expect(
+        matches.length,
+        matches.length === 0
+          ? 'unreachable'
+          : `Production bundle contains '${needle}' in ${matches.length} file(s) — this would leak ` +
+              `the Plan 01-11 test-hook surface to production. The Vite MODE-gate on the dynamic ` +
+              `import has regressed (verify the literal-comparison branch in src/background/index.ts ` +
+              `or src/offscreen/recorder.ts is still on the static-replacement path). Offending files:\n` +
+              matches
+                .map((m) => `  - ${m.filePath} (${m.count} occurrence${m.count === 1 ? '' : 's'})`)
+                .join('\n'),
+      ).toBe(0);
+    });
+  }
+});